Home

PostgreSQL 7.2 Developer`s Guide

1. which results in two directory trees RELEASE pgsql and CURRENT pgsql From that point on CVS will keep track of which repository branch is in which directory tree and will allow independent updates of either tree If you are only working on the CURRENT source tree you just do everything as before we started tagging release branches After you ve done the initial checkout on a branch cvs checkout r REL6_4 anything you do within that directory structure is restricted to that branch If you apply a patch to that directory structure and do a cvs commit while inside of it the patch is applied to the branch and only the branch 63 Appendix A The CVS Repository A 3 Getting The Source Via CVSup An alternative to using anonymous CVS for retrieving the PostgreSQL source tree is CVSup CVSup was developed by John Polstra lt jdp polstra com gt to distribute CVS repositories and other file trees for the FreeBSD project A major advantage to using CVSup is that it can reliably replicate the entire CVS repository on your local system allowing fast local access to cvs operations such as log and diff Other advantages include fast synchronization to the PostgreSQL server due to an efficient streaming transfer protocol which only sends the changes since the last update A 3 1 Preparing A CVSup Client System Two directory areas are required for CVSup to do it s job a local CVS repository or s
2. Bytel Q Identifies the message as a query String The query string itself ReadyForQuery B Bytel Z Identifies the message type ReadyForQuery is sent whenever the backend is ready for a new query cycle RowDescription B Bytel T Identifies the message as a row description 46 Chapter 4 Frontend Backend Protocol Int16 Specifies the number of fields in a row may be zero Then for each field there is the following String Specifies the field name Int32 Specifies the object ID of the field type Int16 Specifies the type size Int32 Specifies the type modifier SSLRequest F Int32 8 The size of the packet in bytes Int32 80877103 The SSL request code The value is chosen to contain 1234 in the most significant 16 bits and 5679 in the least 16 significant bits To avoid confusion this code must not be the same as any protocol version number StartupPacket F Int32 296 The size of the packet in bytes Int32 The protocol version number The most significant 16 bits are the major version number The least 16 significant bits are the minor version number LimString64 The database name defaults to the user name if empty LimString32 The user name LimString64 Any additional command line arguments to be passed to the backend child process by the server LimString64 Unused 47 Chapter 4 Frontend Backend Protocol LimString64 The optional tty the ba
3. If the original string contains a linguistic mistake report that or fix it yourself in the program source and translate normally The corrected string can be merged in when the program sources have been updated If the original string contains a factual mistake report that or fix it yourself and do not translate it Instead you may mark the string with a comment in the PO file e Maintain the style and tone of the original string Specifically messages that are not sentences cannot open file Sal should probably not start with a capital letter if your language distin guishes letter case or end with a period if your language uses punctuation marks e If you don t know what a message means or if it is ambiguous ask on the developers mailing list Chances are that English speaking end users might also not understand it or find it ambiguous so it s best to improve the message 58 Chapter 9 Native Language Support 9 2 For the Programmer This section describes how to support native language support in a program or library that is part of the PostgreSQL distribution Currently it only applies to C programs Adding NLS support to a program 1 Insert this code into the startup sequence of the program ifdef ENABLE_NLS include lt locale h gt endif ifdef ENABLE_NLS setlocale LC_ALL bindtextdomain progname LOCALEDIR textdomain progname endif The progn
4. AuthenticationKerberosV4 B Bytel R Identifies the message as an authentication request Int32 1 Specifies that Kerberos V4 authentication is required AuthenticationKerberosV5 B Bytel R Identifies the message as an authentication request Int32 2 Specifies that Kerberos V5 authentication is required AuthenticationCleartextPassword B Bytel R Identifies the message as an authentication request 41 Chapter 4 Frontend Backend Protocol Int32 3 Specifies that a cleartext password is required AuthenticationCryptPassword B Bytel R Identifies the message as an authentication request Int32 4 Specifies that a crypt encrypted password is required Byte2 The salt to use when encrypting the password AuthenticationMD5Password B Bytel R Identifies the message as an authentication request Int32 5 Specifies that an MD5 encrypted password is required Byte4 The salt to use when encrypting the password AuthenticationSCMCredential B Bytel R Identifies the message as an authentication request Int32 6 Specifies that an SCM credentials message is required BackendKeyData B Bytel K Identifies the message as cancellation key data The frontend must save these values if it wishes to be able to issue CancelRequest messages later Int32 The process ID of this backend Int32 The secret key of this backend 42 Chapter 4 Frontend Backen
5. Name Type References Description typalign char typalign is the alignment required when storing a value of this type It applies to storage on disk as well as most representations of the value inside PostgreSQL When multiple values are stored consecutively such as in the representation of a complete row on disk padding is inserted before a datum of this type so that it begins on the specified boundary The alignment reference is the beginning of the first datum in the sequence Possible values are e cz CHAR align ment i e no align ment needed e si SHORT alignment 2 bytes on most machines e INT align ment 4 bytes on most machines e d DOUBLE alignment 8 bytes on many machines but by no means all Note For types used in system tables it is critical that the size and alignment defined iN pg_type agree with the way that the compiler will lay out the field in a struct representing a table row 32 Chapter 3 System Catalogs Name Type References Description typstorage char typstorage tells for variable length types those with typlen l if the type is prepared for toasting and what the default strategy for attributes of this type should be Possible values are p Value must always be stored plain e ei Value can be stored in a sec ondary relation if relation has one see pg_class reltoast relid e m Value c
6. e For the RTF version of the same doc src sgml gmake admin rtf e To get a DVI version via JadeTex doc src sgml gmake admin dvi e And Postscript from the DVI doc src sgml gmake admin ps Note The official Postscript format documentation is generated differently See Section B 3 3 below The other books can be built with analogous commands by replacing admin with one of developer programmer tutorial or user Using postgres builds an integrated version of all 5 books which is practical since the browser interface makes it easy to move around all of the documentation by just clicking B 3 1 HTML When building HTML documentation in doc src sgml some of the resulting files will possibly or quite certainly have conflicting names between books Therefore the files are not in that directory in the regular distribution Instead the files belonging to each book are stored in a tar archive that is unpacked at installation time To create a set of HTML documentation packages use the commands cd doc sre gmake tutorial tar gz gmake user tar gz gmake admin tar gz gmake programmer tar gz gmake postgres tar gz gmake install In the distribution these archives live in the doc directory and are installed by default with gmake install 74 Appendix B Documentation B 3 2 Manpages We use the docbook2man utility to convert DocBook REFENTRY pages to roff output suitable for man pages The man pages are als
7. select s sname from supplier s sells se part p where s sno se sno and p pno se pno and s sname lt gt Smith 2 5 Planner Optimizer The task of the planner optimizer is to create an optimal execution plan It first combines all possible ways of scanning and joining the relations that appear in a query All the created paths lead to the same result and it s the task of the optimizer to estimate the cost of executing each path and find out which one is the cheapest 2 5 1 Generating Possible Plans The planner optimizer decides which plans should be generated based upon the types of indexes defined on the relations appearing in a query There is always the possibility of performing a sequential scan on a relation so a plan using only sequential scans is always created Assume an index is defined on a relation for example a B tree index and a query contains the restriction relation attribute OPR constant If relation attribute happens to match the key of the B tree index and OPR is anything but lt gt another plan is created using the B tree index to scan the relation If there are further indexes present and the restrictions in the query happen to match a key of an index further plans will be considered Chapter 2 Overview of PostgreSQL Internals After all feasible plans have been found for scanning single relations plans for joining relations are created The planner optimizer considers only joins between
8. By default only gettext calls are known If you used _ or other identifiers you need to list them here If the translatable string is not the first argument the item needs to be of the form func 2 for the second argument The build system will automatically take care of building and installing the message catalogs To ease the translation of messages here are some guidelines e Do not construct sentences at run time out of laziness like printf Files where s n flag copied removed The word order within the sentence may be different in other languages e For similar reasons this won t work printf copied d filets n n 1 s because it assumes how the plural is formed If you figured you could solve it like this if n 1 printf copied 1 file else printf copied d files n then be disappointed Some languages have more than two forms with some peculiar rules We may have a solution for this in the future but for now this is best avoided altogether You could write printf number of copied files d n e If you want to communicate something to the translator such as about how a message is intended to line up with other output precede the occurrance of the string with a comment that starts with translator g translator This message is not what it seems to be These comments are copied to the message catalog files so that the translators can see them 60 Appen
9. Installing the DocBook DTD Kit 1 Obtain the DocBook V3 1 distribution 2 Create the directory usr local share sgml docbook31 and change to it The exact loca tion is irrelevant but this one is reasonable within the layout we are following here mkdir usr local share sgm1 docbook31 ed usr local share sgml1 docbook31 3 Unpack the archive unzip a docbk31 zip The archive will unpack its files into the current directory 4 Edit the file usr local share sgml catalog or whatever you told jade during installa tion and put a line like this into it CATALOG docbook31 docbook cat 5 Optionally you can edit the file docbook cat and comment out or remove the line containing DTDDECL If you do not then you will get warnings from jade but there is no further harm 6 Download the ISO 8879 character entities archive unpack it and put the files in the same directory you put the DocBook files in ed usr local share sgml1 docbook31 unzip ISOEnts zip 7 Run the following command in the directory with the DocBook and ISO files perl pi e s iso gml1 ISO 1 g docbook cat This fixes a mixup between the names used in the DocBook catalog file and the actual names of the ISO character entity files 12 http www oasis open org docbook sgml 3 1 docbk31 zip 13 http www oasis open org cover ISOEnts zip 72 Appendix B Documentation B 2 4 3 Installing the DocBook DSSSL Sty
10. That s why we find a Sort node in each subplan The additional qualification given in the query s sno gt 2 is pushed down as far as possible and is attached to the qpqua1 field of the leaf SeqScan node of the corresponding subplan The list attached to the field mergeclauses of the MergeJoin node contains information about the join attributes The values 65000 and 65001 for the varno fields in the VAR nodes appearing in the mergeclauses list and also in the target list mean that not the tuples of the current node should be considered but the tuples of the next deeper nodes i e the top nodes of the subplans should be used instead Note that every Sort and SeqScan node appearing in figure ref plan has got a targetlist but because there was not enough space only the one for the MergeJoin node could be drawn Another task performed by the planner optimizer is fixing the operator ids in the Expr and Oper nodes As mentioned earlier PostgreSQL supports a variety of different data types and even user defined types can be used To be able to maintain the huge amount of functions and operators it is necessary to store them in a system table Each function and operator gets a unique operator id According to the types of the attributes used within the qualifications etc the appropriate operator ids have to be used 2 6 Executor The executor takes the plan handed back by the planner optimizer and starts processing the top node In the
11. dropped lanpltrusted bool This is a trusted language See under CREATE LANGUAGE what this means If this is an internal language Lanisp1 is false then this field is meaningless 20 Chapter 3 System Catalogs Name Type References Description lanplcallfoid oid Ipg_proc oid For non internal languages this references the language handler which is a special function that is responsible for executing all functions that are written in the particular language llancompiler text not currently used 3 12 pg _largeobject pg_largeobject holds the data making up large objects A large object is identified by an OID assigned when it is created Each large object is broken into segments or pages small enough to be conveniently stored as rows in pg_largeobject The amount of data per page is defined to be LOBLKSIZE which is currently BLCKSZ 4 or typically 2Kbytes Table 3 12 pg_largeobject Columns Name Type References Description loid oid Identifier of the large object that includes this page pageno int4 Page number of this page within its large object counting from zero data bytea Actual data stored in the large object This will never be more than LOBLKSIZE bytes and may be less Each row of pg_largeobject holds data for one page of a large object beginning at byte offset pageno LOBLKSIZE within the object The im
12. 1 shows how pages in both normal PostgreSQL tables and PostgreSQL indexes e g a B tree index are structured Table 7 1 Sample Page Layout Item Description itemPointerData filler itemData Unallocated Space ItemContinuationData Special Space ItemData 2 ItemData 1 ItemIdData IPageHeaderData The first 8 bytes of each page consists of a page header PageHeaderData Within the header the first three 2 byte integer fields lower upper and special represent byte offsets to the start of unallocated space to the end of unallocated space and to the start of special space Special space is a region at the end of the page that is allocated at page initialization time and contains information specific to an access method The last 2 bytes of the page header opaque encode the page size and information on the internal fragmentation of the page Page size is stored in each page because frames in the buffer pool may be subdivided into equal sized pages on a frame by frame basis within a table The internal fragmentation information is used to aid in determining when page reorganization should occur Following the page header are item identifiers temIdData New item identifiers are allocated from the first four bytes of unallocated space Because an item identifier is never moved until it is freed its index may be used to indicate the location of an item on a page
13. B Documentation B 2 2 FreeBSD Installation The FreeBSD Documentation Project is itself a heavy user of DocBook so it comes as no surprise that there is a full set of ports of the documentation tools available on FreeBSD The following ports need to be installed to build the documentation on FreeBSD e textproc sp e textproc openjade e textproc docbook 310 e textproc iso8879 textproc dsssl docbook modular A number of things from usr ports print tex jadetex might also be of interest It s possible that the ports do not update the main catalog file in usr local share sgml catalog Be sure to have the following line in there CATALOG usr local share sgml docbook 3 1 catalog If you do not want to edit the file you can also set the environment variable SGML_CATALOG_FILES to a colon separated list of catalog files such as the one above More information about the FreeBSD documentation tools can be found in the FreeBSD Documenta tion Project s instructions B 2 3 Debian Packages There is a full set of packages of the documentation tools available for Debian GNU Linux To install simply use apt get install jade apt get install docbook apt get install docbook stylesheets B 2 4 Manual Installation from Source The manual installation process of the DocBook tools is somewhat complex so if you have pre built packages available use them We describe here only a standard setup with reasona
14. Descriptions can be manipulated with the COMMENT command Client applications can view the descriptions by joining with this table Many builtin system objects have comments associated with them that are shown by psql s d commands Table 3 7 pg_description Columns Name Type References Description lobjoid oid any oid attribute The oid of the object this description pertains to classoid oid Ipg_class oid The oid of the system catalog this object appears in 17 Chapter 3 System Catalogs Name Type References Description lobjsubid int4 For a comment on a table attribute this is the attribute s column number the objoid and classoid refer to the table itself For all other object types this field is presently zero description text Arbitrary text that serves as the description of this object 3 8 pg_group This catalog defines groups and stores what users belong to what groups Groups are created with the CREATE GROUP command Consult the Administrator s Guide for information about user permis sion management Because user and group identities are cluster wide pg_group is shared across all databases of a cluster there is only one copy of pg_group per cluster not one per database Table 3 8 pg_group Columns Name Type References Description igroname name Name of the group igrosysid int4 An arbitra
15. Expression tree in the form of a nodeToString representation for partial index predicate 3 10 pg_inherits This catalog records information about table inheritance hierarchies Table 3 10 pg_inherits Columns Name Type References Description inhrelid loid pp class od This is the reference to the subtable that is it records the fact that the identified table is inherited from some other table 19 Chapter 3 System Catalogs Name Type References Description inhparent oid Ipg_class oid This is the reference to the parent table which the table referenced by inhrelid inherited from inhseqno int4 If there is more than one parent for a subtable multiple inheritance this number tells the order in which the inherited columns are to be arranged The count starts at 1 3 11 pg_ language pg_language registers call interfaces or languages in which you can write functions or stored pro cedures See under CREATE LANGUAGE and in the Programmer s Guide for more information about language handlers Table 3 11 pg language Columns Name Type References Description lanname name Name of the language to be specified when creating a function lanispl bool This is false for internal languages such as SQL and true for dynamically loaded language handler modules It essentially means that if it is true the language may be
16. If the name of an attribute cannot be found an error will be returned and the query processing will be aborted 2 4 The PostgreSQL Rule System PostgreSQL supports a powerful rule system for the specification of views and ambiguous view up dates Originally the PostgreSQL rule system consisted of two implementations e The first one worked using tuple level processing and was implemented deep in the executor The rule system was called whenever an individual tuple had been accessed This implementation was removed in 1995 when the last official release of the PostgreSQL project was transformed into Postgres95 e The second implementation of the rule system is a technique called query rewriting The rewrite system is a module that exists between the parser stage and the planner optimizer This technique is still implemented For information on the syntax and creation of rules in the PostgreSQL system refer to The PostgreSQL User s Guide 2 4 1 The Rewrite System The query rewrite system is a module between the parser stage and the planner optimizer It processes the tree handed back by the parser stage which represents a user query and if there is a rule present that has to be applied to the query it rewrites the tree to an alternate form 2 4 1 1 Techniques To Implement Views Now we will sketch the algorithm of the query rewrite system For better illustration we show how to implement views using rules as an example Let the f
17. In fact every pointer to an item ItemPointer created by PostgreSQL consists of a frame number and an index of an item identifier An item identifier contains a byte offset to the start of an item its length in bytes and a set of attribute bits which affect its interpretation The items themselves are stored in space allocated backwards from the end of unallocated space Usually the items are not interpreted However when the item is too long to be placed on a single page or when fragmentation of the item is desired the item is divided and each piece is handled as distinct items in the following manner The first through the next to last piece are placed in an item continuation structure temContinuationData This structure contains itemPointerData which points to the next piece and the piece itself The last piece is handled normally 52 Chapter 8 Genetic Query Optimization Author Written by Martin Utesch lt utesch aut tu freiberg de gt for the Institute of Automatic Control at the University of Mining and Technology in Freiberg Germany 8 1 Query Handling as a Complex Optimization Problem Among all relational operators the most difficult one to process and optimize is the join The number of alternative plans to answer a query grows exponentially with the number of joins included in it Further optimization effort is caused by the support of a variety of join methods e g nested loop hash join merge join in PostgreSQL to
18. SelectStmt node is found it is transformed to a Query node that will be the top most node of the new data structure Figure ref transformed shows the transformed data structure the part for the transformed where clause is given in figure ref transformed_where because there was not enough space to show all parts in one figure Now a check is made if the relation names in the FROM clause are known to the system For every relation name that is present in the system catalogs a RTE node is created containing the relation name the alias name and the relation id From now on the relation ids are used to refer to the relations given in the query All RTE nodes are collected in the range table entry list that is connected to the field rtable of the Query node If a name of a relation that is not known to the system is detected in the query an error will be returned and the query processing will be aborted Next it is checked if the attribute names used are contained in the relations given in the query For every attribute that is found a TLE node is created holding a pointer to a Resdom node which Chapter 2 Overview of PostgreSQL Internals holds the name of the column and a pointer to a VAR node There are two important numbers in the VAR node The field varno gives the position of the relation containing the current attribute in the range table entry list created above The field varattno gives the position of the attribute within the relation
19. case label ti tweak case indent to match PG custom setq indent tabs mode t make sure we keep tabs when indenting For vi your vimrc or equivalent file should contain the following set tabstop 4 or equivalently from within vi try set ts 4 The text browsing tools more and less can be invoked as more x4 less x4 Chapter 2 Overview of PostgreSQL Internals Author This chapter originally appeared as a part of Enhancement of the ANSI SQL Imple mentation of PostgreSQL Stefan Simkovics Master s Thesis prepared at Vienna University of Technology under the direction of O Univ Prof Dr Georg Gottlob and Univ Ass Mag Katrin Seyr This chapter gives an overview of the internal structure of the backend of PostgreSQL After having read the following sections you should have an idea of how a query is processed Don t expect a detailed description here I think such a description dealing with all data structures and functions used within PostgreSQL would exceed 1000 pages This chapter is intended to help understanding the general control and data flow within the backend from receiving a query to sending the results 2 1 The Path of a Query Here we give a short overview of the stages a query has to pass in order to obtain a result 1 A connection from an application program to the PostgreSQL server has to be established The application program transmits a query to the server and receives the results sent back by
20. case of our example the query given in example ref simple_select the top node is a Merge Join node Chapter 2 Overview of PostgreSQL Internals Before any merge can be done two tuples have to be fetched one from each subplan So the executor recursively calls itself to process the subplans it starts with the subplan attached to leftt ree The new top node the top node of the left subplan is a SeqScan node and again a tuple has to be fetched before the node itself can be processed The executor calls itself recursively another time for the subplan attached to leftt ree of the SeqScan node Now the new top node is a Sort node As a sort has to be done on the whole relation the executor starts fetching tuples from the Sort node s subplan and sorts them into a temporary relation in memory or a file when the Sort node is visited for the first time Further examinations of the Sort node will always return just one tuple from the sorted temporary relation Every time the processing of the Sort node needs a new tuple the executor is recursively called for the SeqScan node attached as subplan The relation internally referenced by the value given in the scanrelid field is scanned for the next tuple If the tuple satisfies the qualification given by the tree attached to qpqual it is handed back otherwise the next tuple is fetched until the qualification is satisfied If the last tuple of the relation has been processed a NULL pointer is ret
21. catalog path will not match your location Hence you might need to turn off local variables setq inhibit local variables t The PostgreSQL distribution includes a parsed DTD definitions file reference ced You may find that when using PSGML a comfortable way of working with these separate files of book parts is to insert a proper DOCTYPE declaration while you re editing them If you are working on this source for instance it is an appendix chapter so you would specify the document as an appendix instance of a DocBook document by making the first line look like this lt doctype appendix PUBLIC OASIS DTD DocBook V3 1 EN gt This means that anything and everything that reads SGML will get it right and I can verify the document with nsgmls s docguide sgml But you need to take out that line before building the entire documentation set B 4 2 Other Emacs modes GNU Emacs ships with a different SGML mode which is not quite as powerful as PSGML but it s less confusing and lighter weight Also it offers syntax highlighting font lock which can be very helpful Norm Walsh offers a major mode specifically for DocBook which also has font lock and a number of features to reduce typing 16 http nwalsh com emacs docbookide index html 78 Bibliography Selected references and readings for SQL and PostgreSQL Some white papers and technical reports from the original POSTGRES development team are av
22. detail to give a better understanding on PostgreSQL s internal control and data structures 2 2 How Connections are Established PostgreSQL is implemented using a simple process per user client server model In this model there is one client process connected to exactly one server process As we don t know per se how many connections will be made we have to use a master process that spawns a new Server process every time a connection is requested This master process is called postmaster and listens at a specified Chapter 2 Overview of PostgreSQL Internals TCP IP port for incoming connections Whenever a request for a connection is detected the postmas ter process spawns a new server process called postgres The server tasks postgres processes communicate with each other using semaphores and shared memory to ensure data integrity through out concurrent data access Figure ref connection illustrates the interaction of the master process postmaster the server process postgres and a client application The client process can either be the psql frontend for interactive SQL queries or any user appli cation implemented using the 1ibpg library Note that applications implemented using ecpg the PostgreSQL embedded SQL preprocessor for C also use this library Once a connection is established the client process can send a query to the backend server The query is transmitted using plain text i e there is no parsing done in the front
23. is language specific Currently prosrc contains the function s C language name link symbol for compiled functions both built in and dynamically loaded For all other language types prosrc contains the function s source text Currently probin is unused except for dynamically loaded C functions for which it gives the name of the shared library file containing the function 24 3 16 pg_relcheck Chapter 3 System Catalogs This system catalog stores CHECK constraints on tables Column constraints are not treated spe cially Every column constraint is equivalent to some table constraint See under CREATE TABLE for more information Table 3 16 pg_relcheck Columns Name Type References Description rcrelid oid Ipg_class oid The table this check constraint is on rename name Constraint name rebin text An internal representation of the constraint expression esre text A human readable representation of the constraint expression Note pg_class relchecks needs to match up with the entries in this table 3 17 pg_rewrite This system catalog stores rewrite rules for tables and views Table 3 17 pg_rewrite Columns Name Type References Description Irulename name Rule name ev_type char Event type that the rule is for 1 SELECT X UPDATE 3 INSERT 4 DELETE Ev Class oid pp class od The table this rule is for ev_a
24. is organized into several books e Tutorial introduction for new users e User s Guide documents the SQL implementation e Reference Manual reference pages for programs and SQL commands e Administrator s Guide installation and server maintenance e Programmer s Guide programming client applications and server extensions Developer s Guide assorted information for developers of PostgreSQL proper All books are available as HTML and Postscript The Reference Manual contains reference entries which are also shipped as man pages HTML documentation and man pages are part of a standard distribution and are installed by default Postscript format documentation is available separately for download B 1 DocBook The documentation sources are written in DocBook which is a markup language superficially similar to HTML Both of these languages are applications of the Standard Generalized Markup Language SGML which is essentially a language for describing other languages In what follows the terms DocBook and SGML are both used but technically they are not interchangeable DocBook allows an author to specify the structure and content of a technical document without wor rying about presentation details A document style defines how that content is rendered into one of several final forms DocBook is maintained by the OASIS group The official DocBook site has good introductory and reference documentation and a complete O Re
25. kind for the Nth slot or NULL if the slot kind does not involve numerical values stavaluesN text Column data values of the appropriate kind for the Nth slot or NULL if the slot kind does not store any data values For datatype independence all column data values are converted to external textual form and stored as TEXT datums 3 20 pg_trigger This system catalog stores triggers on tables See under CREATE TRIGGER for more information Table 3 20 pg_trigger Columns Name Type References Description tgrelid oid po Class od The table this trigger is on tgname name Trigger name need not be unique tgfoid oid Ip amp _proc oid The function to be called tgtype int2 Bitmask identifying trigger conditions 28 Chapter 3 System Catalogs Name Type References Description tgenabled bool True if trigger is enabled not presently checked everywhere it should be so disabling a trigger by setting this false does not work reliably tgisconstraint bool True if trigger is a RI constraint tgconstrname name RI constraint name tgconstrrelid oid po Class od The table referenced by an RI constraint tgdeferrable bool True if deferrable tginitdeferred bool True if initially deferred tgnargs int2 Number of argument strings passed to trigger function tgattr int2vector Currently unused tgargs bytea Argument strings to pa
26. one A utility fixrtf is available in doc src sgml to accomplish these repairs oe cd doc src sgml S fixrtf tutorial rtf cd doc srce sgml fixrtf refentry reference rtf The script adds s0 Normal as the zero th style in the document According to Applix Ware the RTF standard would prohibit adding an implicit zero th style though M Word happens to handle this case For repairing REFENTRY sections the script replaces keepn tags with keep 3 Open a new document in Applix Words and then import the RTF file 4 Generate a new ToC using Applix Ware 75 K 6 10 11 Appendix B Documentation Select the existing ToC lines from the beginning of the first character on the first line to the last character of the last line Build a new ToC using Tools BookBuilding CreateToc Select the first three levels of headers for inclusion in the ToC This will replace the existing lines imported in the RTF with a native ApplixWare ToC Adjust the ToC formatting by using Format Style selecting each of the three ToC styles and adjusting the indents for First and Left Use the following values Table B 1 Indent Formatting for Table of Contents Style First Indent inches Left Indent inches TOC Heading 1 0 4 0 4 TOC Heading 2 0 8 0 8 TOC Heading 3 RE dE Work through the document to Adjust page breaks e Adjust table column widths e Insert figures into the
27. server then uses this and the information in the pg_hba conf file to determine what further authentication information it requires the frontend to send if any and responds to the frontend accordingly The frontend then sends any required authentication information Once the server validates this it responds to the frontend that it is authenticated and sends a message indicating successful start up normal case or failure for example an invalid database name In order to serve multiple clients efficiently the server launches a new backend process for each client This is transparent to the protocol however In the current implementation a new child process is created immediately after an incoming connection is detected When the frontend wishes to disconnect it sends an appropriate packet and closes the connection without waiting for a response from the backend Packets are sent as a data stream The first byte determines what should be expected in the rest of the packet The exceptions are packets sent as part of the startup and authentication exchange which comprise a packet length followed by the packet itself The difference is historical 4 2 Protocol This section describes the message flow There are four different types of flows depending on the state of the connection start up query function call and termination There are also special provisions for notification responses and command cancellation which can occur at
28. the language you wanted you can just start working on that file If you need to start a new translation effort then first run the command gmake init po This will create a file progname pot pot to distinguish it from PO files that are in production What does the T stand for I don t know Copy this file to language po and edit it To make it known that the new language is available also edit the file nls mk and add the language or language and country code to the line that looks like AVAIL_LANGUAGES de fr Other languages may appear of course 1 http lcweb loc gov standards iso639 2 englangn html 2 http www din de gremien nas nabd iso3 166ma codlstp 1 en_listp1 htm 57 Chapter 9 Native Language Support As the underlying program or library changes messages may be changed or added by the program mers In this case you do not need to start from scratch Instead run the command gmake update po which will create a new blank message catalog file the pot file you started with and will merge it with the existing PO files If the merge algorithm is not sure about a particular message it marks it fuzzy as explained above For the case where something went really wrong the old PO file is saved with a po old extension 9 1 4 Editing the PO files The PO files can be edited with a regular text editor The translator should only change the area between the quotes after the msgstr directive
29. the next time lex or yacc is called Note The mentioned transformations and compilations are normally done automatically using the makefiles shipped with the PostgreSQL source distribution A detailed description of yacc or the grammar rules given in gram y would be beyond the scope of this paper There are many books and documents dealing with lex and yacc You should be familiar with yacc before you start to study the grammar given in gram y otherwise you won t understand what happens there Chapter 2 Overview of PostgreSQL Internals For a better understanding of the data structures used in PostgreSQL for the processing of a query we use an example to illustrate the changes made to these data structures in every stage This example contains the following simple query that will be used in various descriptions and figures throughout the following sections The query assumes that the tables given in The Supplier Database have already been defined Example 2 1 A Simple Select select s sname se pno from supplier s sells s where s sno gt 2 and s sno se sno Figure ref parsetree shows the parse tree built by the grammar rules and actions given in gram y for the query given in Example 2 1 without the operator tree for the where clause which is shown in figure ref where_clause because there was not enough space to show both data structures in one figure The top node of the tree is a Select Stmt node For every entry appe
30. CopyOutResponse The backend is ready to copy data from a table to the frontend It then sends a CopyDataRows message and then a CompletedResponse message with a tag of COPY CursorResponse Beginning of the response to a SELECT FETCH INSERT UPDATE or DELETE query In the FETCH case the name of the cursor being fetched from is included in the message Otherwise the message always mentions the blank cursor RowDescription Indicates that rows are about to be returned in response to a SELECT or FETCH query The message contents describe the layout of the rows This will be followed by an AsciiRow or BinaryRow message depending on whether a binary cursor was specified for each row being returned to the frontend EmptyQueryResponse An empty query string was recognized ErrorResponse An error has occurred 36 Chapter 4 Frontend Backend Protocol ReadyForQuery Processing of the query string is complete A separate message is sent to indicate this because the query string may contain multiple SQL commands CompletedResponse marks the end of pro cessing one SQL command not the whole string ReadyForQuery will always be sent whether processing terminates successfully or with an error NoticeResponse A warning message has been issued in relation to the query Notices are in addition to other responses i e the backend will continue processing the command The response to a SELECT or FETCH query normally consists of
31. CryptPassword packet If this is the correct password the server responds with an AuthenticationOk otherwise it responds with an ErrorResponse AuthenticationMD5Password The frontend must then send a PasswordPacket containing the password encrypted via MD5 using the 4 character salt specified in the AuthenticationMDS5Password packet If this is the correct password the server responds with an AuthenticationOk otherwise it responds with an ErrorResponse AuthenticationSCMCredential This method is only possible for local Unix domain connections on platforms that support SCM credential messages The frontend must issue an SCM credential message and then send a single data byte The contents of the data byte are uninteresting it s only used to ensure that the server waits long enough to receive the credential message If the credential is acceptable the server responds with an AuthenticationOk otherwise it responds with an ErrorResponse If the frontend does not support the authentication method requested by the server then it should immediately close the connection After having received AuthenticationOk the frontend should wait for further messages from the server The possible messages from the backend in this phase are BackendKeyData This message provides secret key data that the frontend must save if it wants to be able to is sue cancel requests later The frontend should not respond to this message but should continue li
32. CursorResponse RowDescrip tion zero or more AsciiRow or BinaryRow messages and finally CompletedResponse INSERT UPDATE and DELETE queries produce CursorResponse followed by CompletedResponse COPY to or from the frontend invokes special protocol as mentioned above All other query types normally produce only a CompletedResponse message Since a query string could contain several queries separated by semicolons there might be several such response sequences before the backend finishes processing the query string ReadyForQuery is issued when the entire string has been processed and the backend is ready to accept a new query string If a completely empty no contents other than whitespace query string is received the response is EmptyQueryResponse followed by ReadyForQuery The need to specially distinguish this case is historical In the event of an error ErrorResponse is issued followed by ReadyForQuery All further processing of the query string is aborted by ErrorResponse even if more queries remained in it Note that this may occur partway through the sequence of messages generated by an individual query A frontend must be prepared to accept ErrorResponse and NoticeResponse messages whenever it is expecting any other type of message Actually it is possible for NoticeResponse to arrive even when the frontend is not expecting any kind of message that is the backend is nominally idle In particular the backend can be command
33. Installation from Source essesseeseeeseeseseesesetssresressessrssreseesrseserserssreseesees 71 B 2 4 1 Installing OpenJade AAA 71 B 2 4 2 Installing the DocBook DTD 72 B 2 4 3 Installing the DocBook DSSSL Style Sheets A 73 B 24 4 Installing Jade EE 73 B 3 Building The Documentation cece ceeeeeeeceseeeeecaeceseseeseceeceseseeecaecsaesaeeseseseeeeseaeeaa 73 BiB GER E RN EE 74 B32 Mapa SeSe n n vest Gaia ad eee ee i ee a de eel 74 B 3 3 Hardcopy Generation neroni in SE canedbiesseattoraetues bereede 75 BZ Bam Text lee sciatic teetseasnctecl eal een eee cata ceed Sects eee eee 76 BA Documentation AUthoring s Terei E E E A H RAL Emgesi BSG NEE NEE E ee 77 B42 sOther E 78 Bibliography secsessssdsssscsceesesacesosus seavsundsessesovessaoansens sonpscaacossbasneseestneseoneunovasonadenssedensseessesoessecnseadewavensnses 79 List of Tables 3 1 System Catal oO 2s css eree petear a eh d ove na E aae eae eree Sa E aE ii 9 3 2 pe asgresate E e i rn R E aie A E ea ee 10 3 3 p attrdef Columns ni a E wer E ENEA E E E E 10 3 4 pg_attribute Columns n S e A RENEE EA S E E S E E ni 11 3 Dp class COMMIS ssie pr i EO EE deer eer ay e EEEE E ERESSE SSES 13 3 6 pg_atabase Columns 16 3 7 pS description e EE 17 3 8 pg croup EE 18 3 9 PS NdEX COMMING siessen e a E e aei eaae eies r o e e Ed reck aS 18 3 10 ps ghet Colmer ni aes ih eae le a aie aoe ete ae 19 KA ops language e 20 3 12 pg_largeobject Columns 21 S 13 ups lstener C
34. Ipg_attribute attnum relchecks int2 Number of check constraints on the table see pg_relcheck catalog reltriggers int2 Number of triggers on the table see pg_trigger catalog relukeys int2 unused Not the number of unique keys relfkeys int2 unused Not the number of foreign keys on the table relrefs ENEZ unused relhasoids bool True if we generate an OID for each row of the relation relhaspkey bool True if the table has or once had a primary key relhasrules bool Table has rules see pg_rewrite catalog relhassubclass bool At least one table inherits from this one 15 Chapter 3 System Catalogs Name Type References Description relacl aclitem Access permissions See the descriptions of GRANT and REVOKE for details 3 6 pg_database The pg_database catalog stores information about the available databases Databases are created with the CREATE DATABASE command Consult the Administrator s Guide for details about the meaning of some of the parameters Unlike most system catalogs pg_database is shared across all databases of a cluster there is only one copy of pg_database per cluster not one per database Table 3 6 pg_database Columns Name Type References Description datname name Database name datdba int4 Ipg_shadow usesysid O
35. OMMS msiri e E E ea E SEEE a at poreenepleespes EESE 21 3 14 ps_operator Columns ireira a ih E E a Ea E E A eee 22 3 13 EpS proc Eltere erer Ee Ee deele 23 3 162 ps relCheCk COMMIS uss ege ege SERA dp these oe E E ESE E 25 KOR pe rewfite RUE 25 3 18 ps shadow Columns EE 26 3 19 pe statistic COlMMMNS ee SEENEN NEES EENS EE saves Seven eb eees 27 3 20 pattie per COMMIS EE 28 3 21 ge type COLUMNS ees EES EEEo Eare ET N EEE Ea EEEE ESET ar ETET a ESen 29 7 l Sample Page Layout eege in aena aot ees E E EE E aa EE ees 52 List of Figures 8 1 Structured Diagram of a Genetic Algorithm esesssesessseesseresressseereresserrsreerereseeresreesreseseeseeesee 53 List of Examples ru EEN 4 Chapter 1 PostgreSQL Source Code 1 1 Formatting Source code formatting uses a 4 column tab spacing currently with tabs preserved i e tabs are not expanded to spaces For Emacs add the following or something similar to your emacs initialization file 7 Check for files with a path containing postgres or pgsql setq auto mode alist cons postgres pgsql ch pgsql c mode auto mode alist setq auto mode alist cons postgres pgsql cc pgsql c mode auto mode alist defun pgsql c mode 7 sets up formatting for PostgreSQL C code interactive c mode setq default tab width 4 c set style bsd set c basic offset to 4 plus other stuff c set offset
36. PROVIDE MAINTENANCE SUPPORT UPDATES ENHANCEMENTS OR MODIFICATIONS Table of Contents 1 PostgreSQL Source Code e essessesossosossessosossesossesosoossososoesosoosessoseosesoosesoosessssoesesossessssossesossesossessssse 1 DD Werer D 1 2 Overview of PostgreSQL Internals cccsssssscsscsssssseecescsssssssscssesescsseesscssecssssssssessseesesssesseeses 2 H E The Path ofa Query E 2 2 2 How Connections are Established AAA 2 2 3 NEE 3 PB SD PATS OL es eee as SEY I TA ae Sd Ms SS ag 3 273 2 KEN ee MT 4 2 4 The PostgreSQL Rule System 5 24l The Rewnte Systemsin N E E E A E a E 5 2 4 1 1 Techniques To Implement Views 5 22D ME E E EE 6 2 5 1 Generating Possible Plans 2 0 eseecesceeceeeeesecacesaeceeceeeeseeeeseaeeaeesaecaesneeeseeseees 6 2 52 Data Structure of the Plan ege EUREN DEENEN EEN 7 Ke Ee EE H EKOS 9 SA POVELVIEW iesse ae Aelita EE EE e 9 BZ PE ASOTCOALS 2 ee TTTS 9 SB APR EE 10 Sne EE 11 3 35 EE 13 36 pe database sess ccsscsicisk egene EE SEE EES 16 EE Oe WEE 17 3 8 PSU BLOUP ege SEET ee lesional ane EC ued 18 3 9 PBN sc65 5 Secs sescens EES EENS Seach asennees guaeet seven EE 18 BAO Pe Merits ioc ses sssss EE EEE 19 2 ope langues 20 3 12 pe TarBeObject niisiis Red E S EEAS EE E R EEE EES 21 3 13 pe Stem ens sei e earr asda ae E we EE ee ee teas 21 BVA E 22 SAD x PEs EE 23 SB LOs pe TElCheck neria re avekieg iavissesaneasaies SERA E E E eee ASES 24 3 17 pE TEWE EE Dee ake dl Mi E E Gaeta Se
37. PostgreSQL 7 2 Developer s Guide The PostgreSQL Global Development Group PostgreSQL 7 2 Developer s Guide by The PostgreSQL Global Development Group Copyright 1996 2001 by The PostgreSQL Global Development Group This document contains assorted information that can be of use to PostgreSQL developers Legal Notice PostgreSQL is Copyright 1996 2001 by the PostgreSQL Global Development Group and is distributed under the terms of the license of the University of California below Postgres95 is Copyright 1994 5 by the Regents of the University of California Permission to use copy modify and distribute this software and its documentation for any purpose without fee and without a written agreement is hereby granted provided that the above copyright notice and this paragraph and the following two paragraphs appear in all copies IN NO EVENT SHALL THE UNIVERSITY OF CALIFORNIA BE LIABLE TO ANY PARTY FOR DIRECT INDIRECT SPECIAL INCIDENTAL OR CONSEQUENTIAL DAMAGES INCLUDING LOST PROFITS ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION EVEN IF THE UNIVERSITY OF CALIFORNIA HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE THE UNIVERSITY OF CALIFORNIA SPECIFICALLY DISCLAIMS ANY WARRANTIES INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE THE SOFTWARE PRO VIDED HEREUNDER IS ON AN AS IS BASIS AND THE UNIVERSITY OF CALIFORNIA HAS NO OBLIGATIONS TO
38. Specifies the object ID of the function to call Int32 Specifies the number of arguments being supplied to the function Then for each argument there is the following Int32 Specifies the size of the value of the argument excluding this size Byten Specifies the value of the field itself in binary format n is the above size FunctionResultResponse B Bytel V Identifies the message as a function call result Bytel G Specifies that a nonempty result was returned Int32 Specifies the size of the value of the result excluding this size Byten Specifies the value of the result itself in binary format n is the above size Byte1 0 Unused Strictly speaking FunctionResultResponse and FunctionVoidResponse are the same thing but with some optional parts to the message Function VoidResponse B Bytel V Identifies the message as a function call result Byte1 0 Specifies that an empty result was returned 45 Chapter 4 Frontend Backend Protocol NoticeResponse B Bytel N Identifies the message as a notice String The notice message itself NotificationResponse B Bytel A Identifies the message as a notification response Int32 The process ID of the notifying backend process String The name of the condition that the notify has been raised on PasswordPacket F Int32 The size of the packet in bytes String The password encrypted if requested Query F
39. TALL This will create a file INSTALL htm1 that can be saved as text with Netscape Navigator and put into the place of the existing file Netscape seems to offer the best quality for HTML to text conversions over lynx and w3m The file HISTORY can be created similarly using the command gmake HISTORY For the file src test regress README the command is gmake regress_README B 4 Documentation Authoring SGML and DocBook do not suffer from an oversupply of open source authoring tools The most common toolset is the Emacs XEmacs editor with appropriate editing mode On some systems these tools are provided in a typical full installation B 4 1 Emacs PSGML PSGML is the most common and most powerful mode for editing SGML documents When properly configured it will allow you to use Emacs to insert tags and check markup consistency You could use it for HTML as well Check the PSGML web site for downloads installation instructions and detailed documentation There is one important thing to note with PSGML its author assumed that your main SGML DTD directory would be usr local lib sgmi If as in the examples in this chapter you use usr local share sgml you have to compensate for this either by setting SGML_CATALOG_FILES environment variable or you can customize your PSGML installation its manual tells you how Put the following in your emacs environment file adjusting the path names to be appropriate for your sys
40. ail able at the University of California Berkeley Computer Science Department web site SQL Reference Books Judith Bowman Sandra Emerson and Marcy Darnovsky The Practical SQL Handbook Using Struc tured Query Language Third Edition Addison Wesley ISBN 0 201 44787 8 1996 C J Date and Hugh Darwen A Guide to the SQL Standard A user s guide to the standard database language SQL Fourth Edition Addison Wesley ISBN 0 201 96426 0 1997 C J Date An Introduction to Database Systems Volume 1 Sixth Edition Addison Wesley 1994 Ramez Elmasri and Shamkant Navathe Fundamentals of Database Systems 3rd Edition Addison Wesley ISBN 0 805 31755 4 August 1999 Jim Melton and Alan R Simon Understanding the New SQL A complete guide Morgan Kaufmann ISBN 1 55860 245 3 1993 Jeffrey D Ullman Principles of Database and Knowledge Base Systems Volume 1 Computer Sci ence Press 1988 PostgreSQL Specific Documentation Stefan Simkovics Enhancement of the ANSI SQL Implementation of PostgreSQL Department of Information Systems Vienna University of Technology November 29 1998 Discusses SQL history and syntax and describes the addition of INTERSECT and EXCEPT con structs into PostgreSQL Prepared as a Master s Thesis with the support of O Univ Prof Dr Georg Gottlob and Univ Ass Mag Katrin Seyr at Vienna University of Technology A Yu and J Chen The POSTGRES Group The Postgres95 User Ma
41. all this is that for reasons of both security and efficiency the frontend has no direct way to tell whether a cancel request has succeeded It must continue to wait for the backend to respond to the query Issuing a cancel simply improves the odds that the current query will finish soon and improves the odds that it will fail with an error message instead of succeeding Since the cancel request is sent across a new connection to the server and not across the regular frontend backend communication link it is possible for the cancel request to be issued by any process not just the frontend whose query is to be canceled This may have some benefits of flexibility in building multiple process applications It also introduces a security risk in that unauthorized persons might try to cancel queries The security risk is addressed by requiring a dynamically generated secret key to be supplied in cancel requests 4 2 6 Termination The normal graceful termination procedure is that the frontend sends a Terminate message and im mediately closes the connection On receipt of the message the backend immediately closes the con nection and terminates An ungraceful termination may occur due to software failure i e core dump at either end If either frontend or backend sees an unexpected closure of the connection it should clean up and terminate The frontend has the option of launching a new backend by recontacting the server if it doesn t want to t
42. ame can actually be chosen freely Wherever a message that is a candidate for translation is found a call to gettext needs to be inserted E g fprintf stderr panic level d n lvl would be changed to fprintf stderr gettext panic level d n lvl gettext is defined as a no op if no NLS is configured This may tend to add a lot of clutter One common shortcut is to define _ x gettext x Another solution is feasible if the program does much of its communication through one or a few functions such as elog in the backend Then you make this function call gettext internally on all input values Add a file nls mk in the directory with the program sources This file will be read as a makefile The following variable assignments need to be made here CATALOG_NAME The program name as provided in the textdomain call AVAIL LANGUAGES List of provided translations empty in the beginning GETTEXT_FILES List of files that contain translatable strings i e those marked with gettext or an alterna tive solution Eventually this will include nearly all source files of the program If this list gets too long you can make the first file be a and the second word be a file that contains one file name per line 59 Chapter 9 Native Language Support GETTEXT_TRIGGERS The tools that generate message catalogs for the translators to work on need to know what function calls contain translatable strings
43. an be stored compressed inline e xX Value can be stored com pressed inline or in secondary Note that m fields can also be moved out to secondary storage but only as a last resort Ce and x fields are moved first typdefault text typdefault is NULL for types without a default value If it s not INULL it contains the external string representation of the type s default value 33 Chapter 4 Frontend Backend Protocol Note Written by Phil Thompson lt phil river bank demon co uk gt Updates for protocol 2 0 by Tom Lane lt tgl sss pgh pa us gt PostgreSQL uses a message based protocol for communication between frontends and backends The protocol is implemented over TCP IP and also on Unix domain sockets PostgreSQL 6 3 introduced version numbers into the protocol This was done in such a way as to still allow connections from earlier versions of frontends but this document does not cover the protocol used by those earlier versions This document describes version 2 0 of the protocol implemented in PostgreSQL 6 4 and later Higher level features built on this protocol for example how libpq passes certain environment vari ables after the connection is established are covered elsewhere 4 1 Overview A frontend opens a connection to the server and sends a start up packet This includes the names of the user and of the database the user wants to connect to The
44. any time after the start up phase 4 2 1 Start up Initially the frontend sends a StartupPacket The server uses this info and the contents of the pg_hba conf file to determine what authentication method the frontend must use The server then 34 Chapter 4 Frontend Backend Protocol responds with one of the following messages ErrorResponse The server then immediately closes the connection AuthenticationOk The authentication exchange is completed AuthenticationKerberos V4 The frontend must then take part in a Kerberos V4 authentication dialog not described here part of the Kerberos specification with the server If this is successful the server responds with an AuthenticationOk otherwise it responds with an ErrorResponse AuthenticationKerberos V5 The frontend must then take part in a Kerberos V5 authentication dialog not described here part of the Kerberos specification with the server If this is successful the server responds with an AuthenticationOk otherwise it responds with an ErrorResponse AuthenticationCleartextPassword The frontend must then send a PasswordPacket containing the password in clear text form If this is the correct password the server responds with an AuthenticationOk otherwise it responds with an ErrorResponse AuthenticationCryptPassword The frontend must then send a PasswordPacket containing the password encrypted via crypt 3 using the 2 character salt specified in the Authentication
45. aring in the from clause of the SQL query a RangeVar node is created holding the name of the alias and a pointer to a RelExpr node holding the name of the relation All RangeVar nodes are collected in a list which is attached to the field fromClause of the Select Stmt node For every entry appearing in the select list of the SQL query a ResTarget node is created holding a pointer to an Attr node The Attr node holds the relation name of the entry and a pointer to a Value node holding the name of the attribute All ResTarget nodes are collected to a list which is connected to the field targetList of the Select Stmt node Figure ref where_clause shows the operator tree built for the where clause of the SQL query given in Example 2 1 which is attached to the field qual of the Select Stmt node The top node of the operator tree is an A_Expr node representing an AND operation This node has two successors called lexpr and rexpr pointing to two subtrees The subtree attached to lexpr represents the qualification s sno gt 2and the one attached to rexpr represents s sno se sno For every attribute an Attr node is created holding the name of the relation and a pointer to a Value node holding the name of the attribute For the constant term appearing in the query a Const node is created holding the value 2 3 2 Transformation Process The transformation process takes the tree handed back by the parser as input and steps recursively through it If a
46. bley standard installation paths and no fancy features For details you should study the documentation of the respective package and read SGML introductory material B 2 4 1 Installing OpenJade 1 The installation of OpenJade offers a GNU style configure make make install build process Details can be found in the OpenJade source distribution In a nutshell configur nable default catalog usr local share sgml catalog make make install 11 http www freebsd org doc en_US ISO8859 1 books fdp primer tools html 71 Appendix B Documentation Be sure to remember where you put the default catalog you will need it below You can also leave it off but then you will have to set the environment variable SGML_CATALOG_FILES to point to the file whenever you use jade later on This method is also an option if OpenJade is already installed and you want to install the rest of the toolchain locally 2 Additionally you should install the files dsssl dtd fot dtd style sheet dtd and cat alog from the dsss1 directory somewhere perhaps into usr local share sgml dsssl It s probably easiest to copy the entire directory cp R dsssl usr local share sgml 3 Finally create the file usr local share sgm1 catalog and add this line to it CATALOG dsssl catalog This is a relative path reference to the file installed in step 2 Be sure to adjust it if you chose your installation layout differently B 2 4 2
47. call Notices are in addition to other responses i e the backend will continue processing the command A frontend must be prepared to accept ErrorResponse and NoticeResponse messages whenever it is expecting any other type of message Also if it issues any LISTEN commands then it must be prepared to accept NotificationResponse messages at any time see below 4 2 4 Notification Responses If a frontend issues a LISTEN command then the backend will send a NotificationResponse message not to be confused with NoticeResponse whenever a NOTIFY command is executed for the same notification name Notification responses are permitted at any point in the protocol after start up except within another backend message Thus the frontend must be prepared to recognize a NotificationResponse mes sage whenever it is expecting any message Indeed it should be able to handle NotificationResponse messages even when it is not engaged in a query NotificationResponse A NOTIFY command has been executed for a name for which a previous LISTEN command was executed Notifications may be sent at any time It may be worth pointing out that the names used in listen and notify commands need not have any thing to do with names of relations tables in the SQL database Notification names are simply arbi trarily chosen condition names 4 2 5 Cancelling Requests in Progress During the processing of a query the frontend may request cancellation of the qu
48. ce are represented by chromosomes in essence a set of character strings A gene is a subsection of a chromosome which encodes the value of a single parameter being optimized Typical encodings for a gene could be binary or integer Through simulation of the evolutionary operations recombination mutation and selection new gen erations of search points are found that show a higher average fitness than their ancestors According to the comp ai genetic FAQ it cannot be stressed too strongly that a GA is not a pure random search for a solution to a problem A GA uses stochastic processes but the result is distinctly non random better than random Figure 8 1 Structured Diagram of a Genetic Algorithm 53 Chapter 8 Genetic Query Optimization P t generation of ancestors at a time t P O generation of descendants at a time t gt gt gt gt gt gt gt gt gt gt gt Algorithm GA lt lt lt lt lt lt lt lt lt lt lt lt lt lt INITIALIZE t 0 INITIALIZE P t evaluate FITNESS of P t while not STOPPING CRITERION do P t RECOMBINATION P t P t MUTATION P t P t 1 SELECTION P t P t evaluate FITNESS of P t tS at 2 8 3 Genetic Query Optimization GEQO in PostgreSQL The GEQO module is intended for the solution of the query optimization problem similar to a trave
49. ckend should use for debugging messages Currently this field is unsupported and ignored Terminate F Bytel X Identifies the message as a termination 48 Chapter 5 gcc Default Optimizations Note Contributed by Brian Gallew lt geek cmu edu gt Configuring gcc to use certain flags by default is a simple matter of editing the usr local lib gcc lib platform version specs file The format of this file pretty simple The file is broken into sections each of which is three lines long The first line is section_name e g asm The second line is a list of flags and the third line is blank The easiest change to make is to append the desired default flags to the list in the appropriate section As an example let s suppose that I have linux running on a 486 with gcc 2 7 2 installed in the default location In the file usr local lib gec lib i486 linux 2 7 2 specs 13 lines down I find the following section As you can see there aren t any default flags If I always wanted compiles of C code to use m486 fomit frame pointer I would change it to look like If I wanted to be able to generate 386 code for another older linux box lying around I d have to make it look like this This will always omit frame pointers any will build 486 optimized code unless m386 is specified on the command line You can actually do quite a lot of customization with the specs file Always remembe
50. command separator the next token that syntactically cannot belong to the preceding command starts a new one Usually you would put a new command on a new line for clarity Tokens can be certain key words special characters parentheses commas etc numbers or double quoted strings Everything is case sensitive Lines starting with a are ignored 6 2 BKI Commands open tablename Open the table called t ablename for further manipulation close tablename Close the open table called tablename It is an error if tablename is not already opened If no tablename is given then the currently open table is closed create tablename namel typel name2 typed Create a table named tablename with the columns given in parentheses The type is not necessarily the data type that the column will have in the SQL environment that is determined by the pg_attribute system catalog The type here is essentially only used to allocate storage The following types are allowed bool bytea char 1 byte name int2 int2vector int4 regproc text oid tid xid cid oidvector smgr _int4 array _aclitem array Array types can also be indicated by writing after the name of the element type Note The table will only be created on disk it will not automatically be registered in the system catalogs and will therefore not be accessible unless appropriate rows are inserted in pg_class pg_attribute etc 50 Chapter 6 BKI Backend Int
51. d For an INSERT command the tag is INSERT oid rows where rows is the number of rows inserted and oid is the object ID of the inserted row if rows is 1 otherwise oid is 0 43 Chapter 4 Frontend Backend Protocol For a DELETE command the tag is DELE deleted H H E rows where rows is the number of rows For an UPDATE command the tag is UPDATE rows where rows is the number of rows updated CopyDataRows B amp F This is a stream of rows where each row is terminated by a Byte1 n This is then followed by the sequence Bytel Bytel Bytel n CopyInResponse B Bytel G Identifies the message as a Start Copy In response The frontend must now send a Copy DataRows message CopyOutResponse B Bytel H Identifies the message as a Start Copy Out response This message will be followed by a CopyDataRows message CursorResponse B Bytel P Identifies the message as a cursor response String The name of the cursor This will be blank if the cursor is implicit EmptyQueryResponse B Bytel T Identifies the message as a response to an empty query string String Unused ErrorResponse B Bytel E Identifies the message as an error String The error message itself 44 Chapter 4 Frontend Backend Protocol FunctionCall F Bytel F Identifies the message as a function call String Unused Int32
52. d Protocol BinaryRow B Bytel B Identifies the message as a binary data row A prior RowDescription message defines the number of fields in the row and their data types Byten A bit map with one bit for each field in the row The Ist field corresponds to bit 7 MSB of the Ist byte the 2nd field corresponds to bit 6 of the Ist byte the 8th field corresponds to bit 0 LSB of the Ist byte the 9th field corresponds to bit 7 of the 2nd byte and so on Each bit is set if the value of the corresponding field is not NULL If the number of fields is not a multiple of 8 the remainder of the last byte in the bit map is wasted Then for each field with a non NULL value there is the following Int32 Specifies the size of the value of the field excluding this size Byten Specifies the value of the field itself in binary format n is the above size CancelRequest F Int32 16 The size of the packet in bytes Int32 80877102 The cancel request code The value is chosen to contain 1234 in the most significant 16 bits and 5678 in the least 16 significant bits To avoid confusion this code must not be the same as any protocol version number Int32 The process ID of the target backend Int32 The secret key for the target backend CompletedResponse B Bytel C Identifies the message as a completed response String The command tag This is usually a single word that identifies which SQL command was complete
53. de base originally developed by the DEC Systems Research Center The PM3 RPM distribution is roughly 30MB compressed At the time of writing the 1 1 10 1 release installed cleanly on RH 5 2 whereas the 1 1 11 1 release is apparently built for another release RH 6 0 and does not run on RH 5 2 Tip This particular rpm packaging has many RPM files so you will likely want to place them into a separate directory b Install the Modula 3 rpms rpm Uvh pm3 rpm 2 Unpack the cvsup distribution cd usr local sre tar zxf cvsup 16 0 tar gz 7 8 http m3 polymtl ca m3 http www research digital com SRC modula 3 html home html 67 Appendix A The CVS Repository 3 Build the cvsup distribution suppressing the GUI interface feature to avoid requiring X11 li braries make M3FLAGS DNOGUI and if you want to build a static binary to move to systems that may not have Modula 3 installed try make M3FLAGS DNOGUI DSTATIC 4 Install the built binary make M3FLAGS DNOGUI DSTATIC install 68 Appendix B Documentation PostgreSQL has four primary documentation formats e Plain text for pre installation information e HTML for on line browsing and reference Postscript for printing e man pages for quick reference Additionally a number of plain text README type files can be found throughout the PostgreSQL source tree documenting various implementation issues The documentation
54. dix A The CVS Repository The PostgreSQL source code is stored and managed using the CVS code management system At least two methods anonymous CVS and CVSup are available to pull the CVS code tree from the PostgreSQL server to your local machine A 1 Getting The Source Via Anonymous CVS If you would like to keep up with the current sources on a regular basis you can fetch them from our CVS server and then use CVS to retrieve updates from time to time Anonymous CVS You will need a local copy of CVS Concurrent Version Control System which you can get from http www cyclic com or any GNU software archive site We currently recommend version 1 10 the most recent at the time of writing Many systems have a recent version of cvs installed by default Do an initial login to the CVS server cvs d pserver anoncvs anoncvs postgresql org projects cvsroot login You will be prompted for a password just press ENTER You should only need to do this once since the password will be saved in cvspass in your home directory Fetch the PostgreSQL sources cvs z3 d pserver anoncvs anoncvs postgresql org projects cvsroot co P pgsql which installs the PostgreSQL sources into a subdirectory pgsq1 of the directory you are cur rently in Note If you have a fast link to the Internet you may not need z3 which instructs CVS to use gzip compression for transferred data But on a modem speed link it s a v
55. document Center each figure on the page using the centering margins button on the Applix Ware toolbar Note Not all documents have figures You can grep the SGML source files for the string graphic to identify those parts of the documentation that may have figures A few figures are replicated in various parts of the documentation Replace the right justified page numbers in the Examples and Figures portions of the ToC with correct values This only takes a few minutes per document Delete the index section from the document if it is empty Regenerate and adjust the table of contents a b c d Select the ToC field Select Tools gt Book Building gt Create Table of Contents Unbind the ToC by selecting Tools gt Field Editing gt Unprotect Delete the first line in the ToC which is an entry for the ToC itself Save the document as native Applix Words format to allow easier last minute editing later Print the document to a file in Postscript format Compress the Postscript file using gzip Place the compressed file into the doc directory 76 Appendix B Documentation B 3 4 Plain Text Files Several files are distributed as plain text for reading during the installation process The INSTALL file corresponds to the chapter in the Administrator s Guide with some minor changes to account for the different context To recreate the file change to the directory doc src sgml and enter gmake INS
56. ed to terminate by its parent process In that case it will send a NoticeResponse before closing the connec tion It is recommended that the frontend check for such asynchronous notices just before issuing any new command Also if the frontend issues any LISTEN commands then it must be prepared to accept Notification Response messages at any time see below Recommended practice is to code frontends in a state machine style that will accept any message type at any time that it could make sense rather than wiring in assumptions about the exact sequence of messages 4 2 3 Function Call A Function Call cycle is initiated by the frontend sending a FunctionCall message to the backend The backend then sends one or more response messages depending on the results of the function call and finally a ReadyForQuery response message ReadyForQuery informs the frontend that it may safely send a new query or function call The possible response messages from the backend are 37 Chapter 4 Frontend Backend Protocol ErrorResponse An error has occurred FunctionResultResponse The function call was executed and returned a result Function VoidResponse The function call was executed and returned no result ReadyForQuery Processing of the function call is complete ReadyForQuery will always be sent whether pro cessing terminates successfully or with an error NoticeResponse A warning message has been issued in relation to the function
57. eeeeessesececeseeceeceececeneeceeeseeeenees 54 8 3 1 Future Implementation Tasks for PostgreSQL GEO 55 8 4 Further Readings isc oies BEER EE ER sot cutee CREE NEEE 55 9 Native Language Support ssessssscsssersesersersssessssesssssessesesessesesseseesssssessssessssesessesesesseseeses 56 9 1 For th Translators EE 56 9 1 1 Requirements eebe e sees tose ESA 56 9 1 2 Conte EE 56 9 1 3 Creating and maintaining message catalogs s eseesesreseeeesseesreerereersreeerersererses 57 Hl Editing the PO filesi dena he e o EEEE needed SEEE REENE 58 9 2 For E 58 A The CVS Repository sssscssssssssrssssresserssssssscssesessessssessssssesesesseseesenesscssesesseeseesessesessesersesersssens 61 A 1 Getting The Source Via Anonymous CNS 61 A 2 CVS Tree Organi eebe Eed eege Ee 62 A 3 Getting The Source Via CV Sup EEN covgedsteseas sth Es ERES RE Oe TETs EESE 63 A 3 1 Preparing A CVSup Client System ee e eee eceseeseecsececeseeeeeeceeeeseseeecaeeaeeaes 64 A 3 2 Running a CVSup Client 64 A33 Installing CV SUD EE 66 A 3 4 Installation from Sources 67 B Documentation essiri iir raina sA NAASE 69 By ls DO CB OOK 2 dee Dee Eeer DEE a E A E 69 e E CEET E 69 B 2 1 Linux RPM Installation eee eeseceneeceecessecenceceecesaeceaceeseeesaeeeaeeeeecsaeeeaeeeseeees 70 P FreeBSD Installation w i dence binedgincsscgevoustueletvcottete dE eor 70 B 2 3 Debian Packages erann a ane heel ee a ne oe a 71 B 2 4 Manual
58. end client The server parses the query creates an execution plan executes the plan and returns the retrieved tuples to the client by transmitting them over the established connection 2 3 The Parser Stage The parser stage consists of two parts e The parser defined in gram y and scan 1 is built using the Unix tools yacc and lex e The transformation process does modifications and augmentations to the data structures returned by the parser 2 3 1 Parser The parser has to check the query string which arrives as plain ASCII text for valid syntax If the syntax is correct a parse tree is built up and handed back otherwise an error is returned For the implementation the well known Unix tools lex and yacc are used The lexer is defined in the file scan 1 and is responsible for recognizing identifiers the SOL keywords etc For every keyword or identifier that is found a token is generated and handed to the parser The parser is defined in the file gram y and consists of a set of grammar rules and actions that are executed whenever a rule is fired The code of the actions which is actually C code is used to build up the parse tree The file scan 1 is transformed to the C source file scan c using the program lex and gram y is transformed to gram c using yacc After these transformations have taken place a normal C compiler can be used to create the parser Never make any changes to the generated C files as they will be overwritten
59. er platform check for and download the appropriate binary from the PostgreSQL ftp site 2 Check the tar file to verify the contents and directory structure if any For the linux tar file at least the static binary and man page is included without any directory packaging a If the binary is in the top level of the tar file then simply unpack the tar file into your target directory cd usr local bin tar zxvf usr local src cvsup 16 0 linux i386 tar gz mv cvsup 1 doc man manl 5 ftp ftp postgresql org pub 6 ftp ftp postgresql org pub 66 Appendix A The CVS Repository b If there is a directory structure in the tar file then unpack the tar file within usr local src and move the binaries into the appropriate location as above 3 Ensure that the new binaries are in your path rehash which cvsup set path path to cvsup path which cvsup usr local bin cvsup A 3 4 Installation from Sources Installing CVSup from sources is not entirely trivial primarily because most systems will need to install a Modula 3 compiler first This compiler is available as Linux RPM FreeBSD package or source code Note A clean source installation of Modula 3 takes roughly 200MB of disk space which shrinks to roughly 50MB of space when the sources are removed Linux installation 1 Install Modula 3 a Pick up the Modula 3 distribution from Polytechnique Montr al who are actively maintaining the co
60. erface insert OID oid_value valuel value2 Insert a new row into the open table using valuel valued etc for its column values and oid_value for its OID If oid_value is zero 0 or the clause is omitted then the next available OID is used NULL values can be specified using the special key word _nu11_ Values containing spaces must be double quoted declare unique index indexname on tablename using amname opclass1 name1 Create an index named indexname on the table named tablename using the amname access method The fields to index are called namel name2 etc and the operator classes to use are opclassi opclass2 etc respectively build indices Build the indices that have previously been declared 6 3 Example The following sequence of commands will create the test_table table with the two columns cola and colb of type int 4 and text respectively and insert two rows into the table create test_table cola int4 colb text open test_table insert OID 421 1 valuel insert OID 422 2 _null_ close test_table 51 Chapter 7 Page Files A description of the database file default page format This section provides an overview of the page format used by PostgreSQL tables User defined access methods need not use this page format In the following explanation a byte is assumed to contain 8 bits In addition the term item refers to data that is stored in PostgreSQL tables Table 7
61. erminate itself For either normal or abnormal termination any open transaction is rolled back not committed One should note however that if a frontend disconnects while a query is being processed the backend will probably finish the query before noticing the disconnection If the query is outside any transaction block BEGIN COMMIT sequence then its results may be committed before the disconnection is recognized 4 2 7 SSL Session Encryption Recent releases of PostgreSQL allow frontend backend communication to be encrypted using SSL This provides communication security in environments where attackers might be able to capture the session traffic To initiate an SSL encrypted connection the frontend initially sends an SSLRequest message rather than a StartupPacket The server then responds with a single byte containing Y or N indicating that it is willing or unwilling to perform SSL respectively The frontend may close the connection at this point if it is dissatisfied with the response To continue after y perform an SSL startup handshake not described here part of the SSL specification with the server If this is successful continue with sending the usual StartupPacket In this case the StartupPacket and all subsequent data will be SSL encrypted To continue after N send the usual StartupPacket and proceed without encryption 39 Chapter 4 Frontend Backend Protocol The frontend should also be prepared to handle an ErrorMe
62. ery The cancel request is not sent directly on the open connection to the backend for reasons of implementation efficiency we don t want to have the backend constantly checking for new input from the frontend during query processing Cancel requests should be relatively infrequent so we make them slightly cumbersome in order to avoid a penalty in the normal case To issue a cancel request the frontend opens a new connection to the server and sends a Cancel Request message rather than the StartupPacket message that would ordinarily be sent across a new 38 Chapter 4 Frontend Backend Protocol connection The server will process this request and then close the connection For security reasons no direct reply is made to the cancel request message A CancelRequest message will be ignored unless it contains the same key data PID and secret key passed to the frontend during connection start up If the request matches the PID and secret key for a currently executing backend the processing of the current query is aborted In the existing implementation this is done by sending a special signal to the backend process that is processing the query The cancellation signal may or may not have any effect for example if it arrives after the backend has finished processing the query then it will have no effect If the cancellation is effective it results in the current command being terminated early with an error message The upshot of
63. ery substantial win This initial checkout is a little slower than simply downloading a tar gz file expect it to take 40 minutes or so if you have a 28 8K modem The advantage of CVS doesn t show up until you want to update the file set later on Whenever you want to update to the latest CVS sources ed into the pgsql subdirectory and issue cvs z3 update d P This will fetch only the changes since the last time you updated You can update in just a couple of minutes typically even over a modem speed line You can save yourself some typing by making a file cvsrc in your home directory that contains cvs z3 update d P 61 Appendix A The CVS Repository This supplies the z3 option to all cvs commands and the a and P options to cvs update Then you just have to say cvs update to update your files Caution Some older versions of CVS have a bug that causes all checked out files to be stored world writable in your directory If you see that this has happened you can do something like chmod R go w pgsql to set the permissions properly This bug is fixed as of CVS version 1 9 28 CVS can do a lot of other things such as fetching prior revisions of the PostgreSQL sources rather than the latest development version For more info consult the manual that comes with CVS or see the online documentation at http www cyclic com A 2 CVS Tree Organization Author Written by Marc G Fournier lt scra
64. es or the language the file is for depending on the situation This is a bit confusing Examples are psql po PO file for psql or fr mo MO file in French The file format of the PO files is illustrated here comment msgid original string msgstr translated string msgid more original msgstr another translated string can be broken up like this 56 Chapter 9 Native Language Support The msgid s are extracted from the program source They need not be but this is the most common way The msgstr lines are initially empty and are filled in with useful strings by the translator The strings can contain C style escape characters and can be continued across lines as illustrated The next line must start at the beginning of the line The character introduces a comment If whitespace immediately follows the character then this is a comment maintained by the translator There may also be automatic comments which have a non whitespace character immediately following the These are maintained by the various tools that operate on the PO files and are intended to aid the translator automatic comment filename c 1023 flags flags The style comments are extracted from the source file where the message is used Possibly the programmer has inserted information for the translator such as about expected alignment The comment indicates the exact location s where the message is used in the source The trans
65. es 25 BUS ipo SHAW EE 26 3 19 pg EE 26 3 20 POTS BOL snn eaa en EEE e E E OES EE R E e 28 SEA EE 29 4 Frontend Backend Protocol scssssscsssssssssssscsssssoscssssssscsssscsssssescssessensscesscsssssssssssssssssenssonss 34 AV HOV E 34 AD PLOLOCO EE 34 ADV Start EE 34 ADD Query EE 36 AD 3 Function EE 37 4 2 4 Notification Responses 0 0 0 0 cecececceeeeeeeeeseecsecaeceaeeseeseceeceseseeecaessaesaeeseeeseeseeeeeees 38 4 2 5 Cancelling Requests in Progress e cesses cseceeceseeseeeeceeeeseeeaecaecaeeseeeseeeeeeeeeeens 38 42 6 Ee E EE 39 4 277 SSE eege BNCLY Pum eserse ereire p ea a EEES SE o PERE SR Eesi 39 4 3 Message Data Types vices ie ie ditt a r E A e ENEE 40 iii at Message Forinats ccna e ti dike he keene ae ote Ansan ones 40 5 gcc Default Optimizations ssssescessrersersrsessrsessssessssesesssesessesessesessesessessssessssesessesesesseserses 49 6 BKI Backend Inter face sssccssrsssersssesecsessesessssessssessssesessesesesseseeseseesessssessssessssesessesesesseserses 50 EE a R File He 6 2 BRE Etage deen cis cedsscess ccc vas ceed Ee gees hava sis bd ee 6 3 Example eegend eh ate ed liens ett aie Stok ahi ER await 7 Page Files uaran S 8 Genetic Query Optimization 8 1 Query Handling as a Complex Optimization Problem 000 eee eeeeseeeeceeceeeeeeeeeeeneenaes 53 8 2 Genetic Algorithms ee REESEN eebe EEEE a EEEE SEESE 53 8 3 Genetic Query Optimization GEQO in PosteresOl
66. every two relations for which there exists a corresponding join clause i e for which a restriction like where rell attrl rel2 attr2 exists in the where qualification All possible plans are generated for every join pair considered by the planner optimizer The three possible join strategies are e nested iteration join The right relation is scanned once for every tuple found in the left relation This strategy is easy to implement but can be very time consuming e merge sort join Each relation is sorted on the join attributes before the join starts Then the two re lations are merged together taking into account that both relations are ordered on the join attributes This kind of join is more attractive because every relation has to be scanned only once e hash join the right relation is first hashed on its join attributes Next the left relation is scanned and the appropriate values of every tuple found are used as hash keys to locate the tuples in the right relation 2 5 2 Data Structure of the Plan Here we will give a little description of the nodes appearing in the plan Figure ref plan shows the plan produced for the query in example ref simple_select The top node of the plan is a MergeJoin node that has two successors one attached to the field lefttree and the second attached to the field righttree Each of the subnodes represents one relation of the join As mentioned above a merge sort join requires each relation to be sorted
67. fault base usr local pgsql 4 ftp ftp postgresql org pub CVSup README cvsup 65 Appendix A The CVS Repository prefix directory where CVSup will store the actual distribution s default prefix usr local pgsql complete distribution including all below pgsql individual distributions vs the whole thing pgsql doc pgsql perl15 pgsql sre A 3 3 Installing CVSup CVSup is available as source pre built binaries or Linux RPMs It is far easier to use a binary than to build from source primarily because the very capable but voluminous Modula 3 compiler is required for the build CVSup Installation from Binaries You can use pre built binaries if you have a platform for which binaries are posted on the PostgreSQL ftp site or if you are running FreeBSD for which CVSup is available as a port Note CVSup was originally developed as a tool for distributing the FreeBSD source tree It is available as a port and for those running FreeBSD if this is not sufficient to tell how to obtain and install it then please contribute a procedure here At the time of writing binaries are available for Alpha Tru64 ix86 xBSD HPPA HP UX 10 20 MIPS IRIX ix86 linux libc5 ix86 linux glibc Sparc Solaris and Sparc SunOS 1 Retrieve the binary tar file for cvsup cvsupd is not required to be a client appropriate for your platform a Ifyou are running FreeBSD install the CVSup port b Ifyou have anoth
68. ferences Description 10 Chapter 3 System Catalogs Name Type References Description ladrelid oid Ipg_class oid The table this column belongs to ladnum int2 Ipg_attribute attnum The number of the column ladbin text An internal representation of the column default value adsrc text A human readable representation of the default value 3 4 pg_attribute pg_attribute stores information about table columns There will be exactly one pg_attribute row for every column in every table in the database There will also be attribute entries for indexes and other objects See pg_class The term attribute is equivalent to column and is used for historical reasons Table 3 4 pg_attribute Columns Name Type References Description attrelid oid po Class od The table this column belongs to lattname name Column name atttypid oid Ipg_type oid The data type of this column lattstattarget ant4 attstattarget controls the level of detail of statistics accumulated for this column by ANALYZE IA zero value indicates that no statistics should be collected The exact meaning of positive values is datatype dependent For scalar datatypes attstattarget is both the target number of most common values to collect and the target number of histogram bins to create Il Chapter 3 System Catalogs Name Type References Description attlen int2 This
69. illy book for your online reading pleasure The FreeBSD Documentation Project also uses DocBook and has some good information including a number of style guidelines that might be worth considering 1 http www oasis open org 2 http www oasis open org docbook 3 http www freebsd org docproj docproj html 69 Appendix B Documentation B 2 Toolsets The following tools are used to process the documentation Some may be optional as noted DocBook DTD This is the definition of DocBook itself We currently use version 3 1 you cannot use later or earlier versions Note that there is also an XML version of DocBook do not use that ISO 8879 character entities These are required by DocBook but are distributed separately because they are maintained by ISO OpenJade This is the base package of SGML processing It contains an SGML parser a DSSSL processor that is a program to convert SGML to other formats using DSSSL stylesheets as well as a number of related tools Jade is now being maintained by the OpenJade group no longer by James Clark DocBook DSSSL Stylesheets These contain the processing instructions for converting the DocBook sources to other formats such as HTML DocBook2X tools This optional package is used to create man pages It has a number of prerequisite packages of its own Check the web site Jade Tex If you want to you can also install JadeTeX to use TeX as a formatting backe
70. imply a di rectory area if you are fetching a snapshot rather than a repository see below and a local CVSup bookkeeping area These can coexist in the same directory tree Decide where you want to keep your local copy of the CVS repository On one of our systems we recently set up arepository in home cvs but had formerly kept it under a PostgreSQL development tree in opt postgres cvs If you intend to keep your repository in home cvs then put setenv CVSROOT home cvs in your cshrc file or a similar line in your bashrc or profile file depending on your shell The cvs repository area must be initialized Once CVSROOT is set then this can be done with a single command cvs init after which you should see at least a directory named CVSROOT when listing the CVSROOT directory ls CVSROOT CVSROOT A 3 2 Running a CVSup Client Verify that cvsup is in your path on most systems you can do this by typing which cvsup Then simply run cvsup using cvsup L 2 postgres cvsup where L 2 enables some status messages so you can monitor the progress of the update and post gres cvsup is the path and name you have given to your CVSup configuration file 3 http www freebsd org 64 Appendix A The CVS Repository Here is a CVSup configuration file modified for a specific installation and which maintains a full local CVS repository This file represents the standard CVSup distribution file for the P
71. is a copy of the pg_type typlen for this column s type lattnum int2 The number of the column Ordinary columns are numbered from 1 up System columns such as oid have arbitrary negative numbers attndims int4 Number of dimensions if the column is an array type otherwise 0 Presently the number of dimensions of an array is not enforced so any nonzero value effectively means it s an array attcacheoff int4 Always 1 in storage but when loaded into a tuple descriptor in memory this may be updated to cache the offset of the attribute within the tuple latttypmod int4 atttypmod records type specific data supplied at table creation time for example the maximum length of a varchar column It is passed to type specific input and output functions as the third argument The value will generally be 1 for types that do not need typmod lattbyval bool A copy of pg_type typbyval of this column s type lattstorage char A copy of IpPg_type typstorage of this column s type 12 Chapter 3 System Catalogs Name Type References Description lattisset Gool If true this attribute is a Set In that case what is really stored in the attribute is the OID of a tuple in the pg_proc catalog The pg_proc tuple contains the query string that defines this set i e the query to run to get the set So the atttypid see above refers to the type returned b
72. k file of this relation relpages int4 Size of the on disk representation of this table in pages size BLCKSZ This is only an estimate used by the planner It is updated by VACUUM ANALYZE and CREATE INDEX reltuples float4 Number of tuples in the table This is only an estimate used by the planner It is updated by VACUUM ANALYZE and CREATE INDEX reltoastrelid oid Ipg_class oid Oid of the TOAST table associated with this table 0 if none The TOAST table stores large attributes out of line in a secondary table reltoastidxid oid Ipg_class oid For a TOAST table the OID of its index 0 if not a TOAST table relhasindex bool True if this is a table and it has or recently had any indexes This is set by CREATE INDEX but not cleared immediately by DROP INDEX VACUUM clears relhasindex if it finds the table has no indexes 14 Chapter 3 System Catalogs Name Type References Description relisshared Gool True if this table is shared across all databases in the cluster Only certain system catalogs such as lpg_database are shared relkind char y ordinary table i index S sequence Pv view s special Pt secondary TOAST table irelnatts int2 Number of user columns in the relation system columns not counted There must be this many corresponding entries in lpg_attribute See also
73. l ing salesman problem TSP Possible query plans are encoded as integer strings Each string repre sents the join order from one relation of the query to the next E g the query tree 2 CV 4 1 is encoded by the integer string 4 1 3 2 which means first join relation 4 and UU then 3 and then 2 where 1 2 3 4 are relation IDs within the PostgreSQL optimizer Parts of the GEQO module are adapted from D Whitley s Genitor algorithm Specific characteristics of the GEQO implementation in PostgreSQL are e Usage of a steady state GA replacement of the least fit individuals in a population not whole generational replacement allows fast convergence towards improved query plans This is essential for query handling with reasonable time e Usage of edge recombination crossover which is especially suited to keep edge losses low for the solution of the TSP by means of a GA e Mutation as genetic operator is deprecated so that no repair mechanisms are needed to generate legal TSP tours 54 Chapter 8 Genetic Query Optimization The GEQO module allows the PostgreSQL query optimizer to support large join queries effectively through non exhaustive search 8 3 1 Future Implementation Tasks for PostgreSQL GEQO Work is still needed to improve the genetic algorithm parameter settings In file backend optimizer geqo geqo_params c routines gimme_pool_size and gimme_number_generations we have to find a comp
74. lator need not look at the program source but he can if there is doubt about the correct translation The comments contain flags that describe the message in some way There are currently two flags fuzzy is set if the message has possibly been outdated because of changes in the program source The translator can then verify this and possibly remove the fuzzy flag Note that fuzzy messages are not made available to the end user The other flag is c format which indicates that the message is a print f style format template This means that the translation should also be a format string with the same number and type of placeholders There are tools that can verify this which key off the c format flag 9 1 3 Creating and maintaining message catalogs Okay so how does one create a blank message catalog First go into the directory that contains the program whose messages you want to translate If there is a file nls mk then this program has been prepared for translation If there are already some po files then someone has already done some translation work The files are named language po where language is the ISO 639 1 two letter language code in lower case e g fr po for French If there is really a need for more than one translation effort per language then the files may also be named language_region po where region is the ISO 3166 1 two letter country code in upper case e g pt_BR po for Portuguese in Brazil If you find
75. le Sheets To install the style sheets unzip and untar the distribution and move it to a suitable place for example usr local share sgml The archive will automatically create a subdirectory gunzip docbook dsssl 1 xx tar gz tar C usr local share sgml xf docbook dsssl 1 xx tar The usual catalog entry in usr local share sgml catalog can also be made CATALOG docbook dsssl 1 xx catalog Because stylesheets change rather often and it s sometimes beneficial to try out alternative versions PostgreSQL doesn t use this catalog entry See Section B 3 for information about how to select the stylesheets instead B 2 4 4 Installing JadeTeX To install and use JadeTeX you will need a working installation of TeX and LaTeX2e including the supported tools and graphics packages Babel AMS fonts and AMS LaTeX the PSNFSS extension and companion kit of the 35 fonts the dvips program for generating PostScript the macro packages fancyhdr hyperref minitoc url and ot2enc All of these can be found on your friendly neighborhood CTAN site The installation of the TeX base system is far beyond the scope of this introduction Binary packages should be available for any system that can run TeX Before you can use JadeTeX with the PostgreSQL documentation sources you will need to increase the size of TeX s internal data structures Details on this can be found in the JadeTeX installation instructions Once that is finished you ca
76. may add comments and alter the fuzzy flag There is unsurprisingly a PO mode for Emacs which I find quite useful The PO files need not be completely filled in The software will automatically fall back to the original string if no translation or an empty translation is available It is no problem to submit incomplete translations for inclusions in the source tree that gives room for other people to pick up your work However you are encouraged to give priority to removing fuzzy entries after doing a merge Re member that fuzzy entries will not be installed they only serve as reference what might be the right translation Here are some things to keep in mind while editing the translations Make sure that if the original ends with a newline the translation does too Similarly for tabs etc If the original is a printf format string the translation also needs to be The translation also needs to have the same format specifiers in the same order Sometimes the natural rules of the language make this impossible or at least awkward In this case you can use this format msgstr Die Datei 2 s hat 1 u Zeichen Then the first placeholder will actually use the second argument from the list The digit ss needs to follow the and come before any other format manipulators This feature really exists in the printf family of functions You may not have heard of it because there is little use for it outside of message internationalization
77. n a more readable format than the underlying pg_statistic table at the cost that its schema must be extended whenever new slot types are added Table 3 19 pg_statistic Columns Name Type References Description starelid oid Ipg_class oid The table that the described column belongs to Staattnum int2 Ipg_attribute attnum The number of the described column stanullfrac float4 The fraction of the column s entries that are NULL stawidth int4 The average stored width in bytes of non NULL entries Stadistinct float4 The number of distinct Inon NULL data values in the column A value greater than zero is the actual number of distinct values A value less than zero is the negative of a fraction of the number of rows in the table for example a column in which values appear about twice on the average could be represented by stadistinct 0 5 A zero value means the number of distinct values is unknown stakindN int2 IA code number indicating the kind of Statistics stored in the Nth slot of the Ipg_statistic row 27 Chapter 3 System Catalogs Name Type References Description staopN Ipg_operator oid An operator used to derive the statistics stored in the Nth slot For example a histogram slot would show the lt operator that defines the sort order of the data stanumbersN float4 Numerical statistics of the appropriate
78. n install JadeTeX gunzip jadetex xxx tar gz tar xf jadetex xxx tar cd jadetex make install mktexlsr Oe Ex Ek Se Ze The last two need to be done as root B 3 Building The Documentation Before you can build the documentation you need to run the configure script as you would when building the programs themselves Check the output near the end of the run it should look something like this checking for onsgmls onsgmls checking for openjade openjade checking for DocBook V3 1 yes checking for DocBook stylesheets usr lib sgml stylesheets nwalsh modular checking for sgmlspl sgmlspl 14 http www ctan org 73 Appendix B Documentation If neither onsgm1s nor nsgm1s were found then you will not see the remaining 4 lines nsgm1s is part of the Jade package If DocBook V3 1 was not found then you did not install the DocBook DTD kit in a place where jade can find it or you have not set up the catalog files correctly See the installation hints above The DocBook stylesheets are looked for in a number of relatively standard places but if you have them some other place then you should set the environment variable DOCBOOKSTYLE to the location and rerun configure afterwards Once you have everything set up change to the directory doc src sgm1 and run one of the following commands Remember to use GNU make e To build the HTML version of the Administrator s Guide doc src sgml gmake admin html
79. nd for Jade JadeTeX can create Postscript or PDF files the latter with bookmarks However the output from JadeTeX is inferior to what you get from the RTF backend Particular problem areas are tables and various artifacts of vertical and horizontal spacing Also there is no opportunity to manually polish the results We have documented experience with several installation methods for the various tools that are needed to process the documentation These will be described below There may be some other packaged distributions for these tools Please report package status to the docs mailing list and we will include that information here B 2 1 Linux RPM Installation Many vendors provide a complete RPM set for DocBook processing in their distribution which is usu ally based on the docbook tools effort at Red Hat Software Look for an SGML option while in stalling or the following packages sgml common docbook stylesheets openjade or jade Possibly sgml tools will be needed as well If your distributor does not provide these then you should be able to make use of the packages from some other reasonably compatible vendor 0 http www oasis open org docbook sgml http www oasis open org cover ISOEnts zip http openjade sourceforge net http docbook sourceforge net projects dsssl index html http docbook2x sourceforge net http jadetex sourceforge net http sources redhat com docbook tools 70 Appendix
80. need anyway To try out your work follow the applicable portions of the installation instructions If you want to start a new translation effort or want to do a message catalog merge described later you will need the programs xgettext and msgmerge respectively in a GNU compatible imple mentation Later we will try to arrange it so that if you use a packaged source distribution you won t need xgettext From CVS you will still need it GNU gettext 0 10 36 or later is currently recommended Your local gettext implementation should come with its own documentation Some of that is probably duplicated in what follows but for additional details you should look there 9 1 2 Concepts The pairs of original English messages and their possibly translated equivalents are kept in mes sage catalogs one for each program although related programs can share a message catalog and for each target language There are two file formats for message catalogs The first is the PO file for Portable Object which is a plain text file with special syntax that translators edit The second is the MO file for Machine Object which is a binary file generated from the respective PO file and is used while the internationalized program is run Translators do not deal with MO files in fact hardly anyone does The extension of the message catalog file is to no surprise either Goor mo The base name is either the name of the program it accompani
81. nual University of California Sept 5 1995 Zelaine Fong The design and implementation of the POSTGRES query optimizer University of California Berkeley Computer Science Department 1 http s2k ftp CS Berkeley EDU 8000 postgres papers 2 http s2k ftp CS Berkeley EDU 8000 postgres papers UCB MS zfong pdf 79 Bibliography Proceedings and Articles Nels Olson Partial indexing in POSTGRES research project University of California UCB Engin T7 49 1993 0676 1993 L Ong and J Goh A Unified Framework for Version Modeling Using Production Rules in a Database System ERL Technical Memorandum M90 33 University of California April 1990 L Rowe and M Stonebraker The POSTGRES data model Proc VLDB Conference Sept 1987 P Seshadri and A Swami Generalized Partial Indexes Proc Eleventh International Conference on Data Engineering 6 10 March 1995 IEEE Computer Society Press Cat No 95CH35724 1995 p 420 7 M Stonebraker and L Rowe The design of POSTGRES Proc ACM SIGMOD Conference on Management of Data May 1986 M Stonebraker E Hanson and C H Hong The design of the POSTGRES rules system Proc IEEE Conference on Data Engineering Feb 1987 M Stonebraker The design of the POSTGRES storage system Proc VLDB Conference Sept 1987 M Stonebraker M Hearst and S Potamianos A commentary on the POSTGRES rules system SIGMOD Record 18 3 Sep
82. o distributed as a tar archive similar to the HTML version To create the man page package use the commands cd doc sre gmake man which will result in a tar file being generated in the doc src directory The man build leaves a lot of confusing output and special care must be taken to produce quality results There is still room for improvement in this area B 3 3 Hardcopy Generation The hardcopy Postscript documentation is generated by converting the SGML source code to RTF then importing into ApplixWare After a little cleanup see the following section the output is printed to a postscript file Several areas are addressed while generating Postscript hardcopy including RTF repair ToC genera tion and page break adjustments Applixware RTF Cleanup jade an integral part of the hardcopy procedure omits specifying a default style for body text In the past this undiagnosed problem led to a long process of Table of Contents ToC generation However with great help from the Applix Ware folks the symptom was diagnosed and a workaround is available 1 Generate the RTF input by typing for example oe cd doc src sgml make tutorial rtf 2 Repair the RTF file to correctly specify all styles in particular the default style If the docu ment contains REFENTRY sections one must also replace formatting hints which tie a preceding paragraph to the current paragraph and instead tie the current paragraph to the following
83. ollowing rule be given create rule view_rul as on select to test_view do instead select s sname p pname from supplier s sells se part p where s sno Se Spo and p pno se pno The given rule will be fired whenever a select against the relation test_view is detected Instead of selecting the tuples from test_view the select statement given in the action part of the rule is executed Chapter 2 Overview of PostgreSQL Internals Let the following user query against test view be given select sname from test_view where sname lt gt Smith Here is a list of the steps performed by the query rewrite system whenever a user query against test_view appears The following listing is a very informal description of the algorithm just in tended for basic understanding For a detailed description refer to A commentary on the POSTGRES rules system test_view Rewrite 1 Take the query given in the action part of the rule 2 Adapt the targetlist to meet the number and order of attributes given in the user query 3 Add the qualification given in the where clause of the user query to the qualification of the query given in the action part of the rule Given the rule definition above the user query will be rewritten to the following form Note that the rewriting is done on the internal representation of the user query handed back by the parser stage but the derived new data structure will represent the following query
84. ostfix right loprisleft bool unused loprcanhash bool This operator supports hash joins loprleft loid Ipg_type oid Type of the left operand loprright oid Ipg_type oid Type of the right operand loprresult oid Ipg_type oid Type of the result loprcom oid Ipg amp _operator oid Commiutator of this operator if any loprnegate oid Ipg_operator oid Negator of this operator if any 22 Chapter 3 System Catalogs Name Type References Description loprlsortop oid Ipg_operator oid If this operator supports merge joins the operator that sorts the type of the left hand operand loprrsortop oid Ipg_operator oid If this operator supports merge joins the operator that sorts the type of the right hand operand loprcode regproc Function that implements this operator loprrest regproc Restriction selectivity estimation function for this operator loprjoin regproc Join selectivity estimation function for this operator 3 15 pg_proc This catalog stores information about functions or procedures The description of CREATE FUNC TION and the Programmer s Guide contain more information about the meaning of some fields Table 3 15 pg_proc Columns Name Type References Description proname name Name of the function proowner int4 Ipg_shadow usesysid Owner creator of the func
85. ostgreSQL ORDBMS project Modified by lockhart fourpalms org 1997 08 28 Point to my local snapshot source tree Pull the full CVS repository not just the latest snapshot Defaults that apply to all the collections defaul defaul lt host cvsup postgresql org Lt default release cvs Lt le compress default delete use rel suffix enable the following line to get the latest snapshot default tag enable the following line to get whatever was specified above or by default at the date specified below default date 97 08 29 00 00 00 base directory where CVSup will store its bookmarks file s will create subdirectory sup default base opt postgres usr local pgsql default base home cvs prefix directory where CVSup will store the actual distribution s default prefix home cvs complete distribution including all below pgsql individual distributions vs the whole thing pgsql doc pgsql perl15 pgsql sre The following is a suggested CVSup config file from the PostgreSQL ftp sitet which will fetch the current snapshot only This file represents the standard CVSup distribution file for the PostgreSQL ORDBMS project Defaults that apply to all the collections default host cvsup postgresql org default compress default release cvs default delete use rel suffix default tag base directory where CVSup will store its bookmarks file s de
86. peruser may not do this unless this attribute is true passwd text Password valuntil labstime Account expiry time only used for password authentication 3 19 pg_statistic pg_statistic stores statistical data about the contents of the database Entries are created by AN ALYZE and subsequently used by the query planner There is one entry for each table column that has been analyzed Note that all the statistical data is inherently approximate even assuming that it is up to date 26 Chapter 3 System Catalogs Since different kinds of statistics may be appropriate for different kinds of data pg_statistic is designed not to assume very much about what sort of statistics it stores Only extremely general statistics such as NULL ness are given dedicated columns in pg_statistic Everything else is stored in slots which are groups of associated columns whose content is identified by a code number in one of the slot s columns For more information see src include catalog pg_statistic h pg_statistic should not be readable by the public since even statistical information about a table s contents may be considered sensitive Example minimum and maximum values of a salary column might be quite interesting pg_stats is a publicly readable view on pg_statistic that only ex poses information about those tables that are readable by the current user pg_stats is also designed to present the information i
87. plementation allows sparse storage pages may be missing and may be shorter than LOBLKSIZE bytes even if they are not the last page of the object Missing regions within a large object read as zeroes 3 13 pg_ listener pg_listener supports the LISTEN and NOTIFY commands A listener creates an entry in pg_listener for each notification name it is listening for A notifier scans pg_listener and updates each matching entry to show that a notification has occurred The notifier also sends a signal using the PID recorded in the table to awaken the listener from sleep 21 Chapter 3 System Catalogs Table 3 13 pg_listener Columns Name Type References Description relname name Notify condition name The name need not match any actual relation in the database the term relname is historical listenerpid int4 PID of the backend process that created this entry notification int4 Zero if no event is pending for this listener If an event is pending the PID of the backend that sent the notification 3 14 pg_operator See CREATE OPERATOR and the Programmer s Guide for details on these operator parameters Table 3 14 pg_operator Columns Name Type References Description loprname name Name of the operator joprowner int4 Ipg_shadow usesysid Owner creator of the operator loprprec int2 unused loprkind char b infix both P prefix left r p
88. ppy hub org gt on 1998 11 05 The command cvs checkout has a flag r that lets you check out a certain revision of a module This flag makes it easy to for example retrieve the sources that make up release 6_4 of the module tc at any time in the future cvs checkout r REL6_4 tc This is useful for instance if someone claims that there is a bug in that release but you cannot find the bug in the current working copy Tip You can also check out a module as it was at any given date using the D option When you tag more than one file with the same tag you can think about the tag as a curve drawn through a matrix of filename vs revision number Say we have 5 files with the following revisions filel file2 file3 file4 filed Lie 1 1 deh Tel che lt TAG 12 e 1 tg N m N 62 Appendix A The CVS Repository then the tag TAG will reference file1 1 2 file2 1 3 etc Note For creating a release branch other then a b option added to the command it s the same thing So to create the 6 4 release I did the following cd pgsql cvs tag b REL6_4 which will create the tag and the branch for the RELEASE tree For those with CVS access it s simple to create directories for different versions First create two subdirectories RELEASE and CURRENT so that you don t mix up the two Then do cd RELEASE cvs checkout P r REL6_4 pgsql cd CURRENT cvs checkout P pgsql
89. process individual joins and a diversity of indexes e g R tree B tree hash in PostgreSQL as access paths for relations The current PostgreSQL optimizer implementation performs a near exhaustive search over the space of alternative strategies This query optimization technique is inadequate to support database applica tion domains that involve the need for extensive queries such as artificial intelligence The Institute of Automatic Control at the University of Mining and Technology in Freiberg Germany encountered the described problems as its folks wanted to take the PostgreSQL DBMS as the backend for a decision support knowledge based system for the maintenance of an electrical power grid The DBMS needed to handle large join queries for the inference machine of the knowledge based system Performance difficulties in exploring the space of possible query plans created the demand for a new optimization technique being developed In the following we propose the implementation of a Genetic Algorithm as an option for the database query optimization problem 8 2 Genetic Algorithms The genetic algorithm GA is a heuristic optimization method which operates through determined randomized search The set of possible solutions for the optimization problem is considered as a population of individuals The degree of adaptation of an individual to its environment is specified by its fitness The coordinates of an individual in the search spa
90. r however that these changes are global and affect all users of the system 49 Chapter 6 BKI Backend Interface Backend Interface BKI files are scripts in a special language that are input to the PostgreSQL back end running in the special bootstrap mode that allows it to perform database functions without a database system already existing BKI files can therefore be used to create the database system in the first place And they are probably not useful for anything else initdb uses a BKI file to do part of its job when creating a new database cluster The input file used by initdb is created as part of building and installing PostgreSQL by a program named genbki sh from some specially formatted C header files in the source tree The created BKI file is called post gres bki and is normally installed in the share subdirectory of the installation tree Related information may be found in the documentation for initdb 6 1 BKI File Format This section describes how the PostgreSQL backend interprets BKI files This description will be easier to understand if the postgres bki file is at hand as an example You should also study the source code of initdb to get an idea of how the backend is invoked BKI input consists of a sequence of commands Commands are made up of a number of tokens depending on the syntax of the command Tokens are usually separated by whitespace but need not be if there is no ambiguity There is no special
91. romise for the parameter settings to satisfy two competing demands e Optimality of the query plan Computing time 8 4 Further Readings The following resources contain additional information about genetic algorithms The Hitch Hiker s Guide to Evolutionary Computation FAQ for comp ai genetic Evolutionary Computation and its application to art and design by Craig Reynolds e Fundamentals of Database Systems e The design and implementation of the POSTGRES query optimizer 1 2 3 http surf de uu net encore www news comp ai genetic http www red3d com cwr evolve html 55 Chapter 9 Native Language Support 9 1 For the Translator PostgreSQL programs server and client can issue their messages in your favorite language if the messages have been translated Creating and maintaining translated message sets needs the help of people who speak their own language well and want to contribute to the PostgreSQL effort You do not have to be a programmer at all to do this This section explains how to help 9 1 1 Requirements We won t judge your language skills this section is about software tools Theoretically you only need a text editor But this is only in the unlikely event that you do not want to try out your translated messages When you configure your source tree be sure to use the enable n1s option This will also check for the libintl library and the msgfmt program which all end users will
92. rue if the type is defined false if this is a placeholder entry for a not yet defined type When typisdefined is false nothing except the type name and OID can be relied on typdelim char Character that separates two values of this type when parsing array input Note that the delimiter is associated with the array element datatype not the array datatype 30 Chapter 3 System Catalogs Name Type References Description typrelid oid po Class od If this is a complex type see typt ype then this field points to the pg_class entry that defines the corresponding table A table could theoretically be used as a composite data type but this is not fully functional typelem oid Ipg_type oid If typelem is not 0 then it identifies another row in pg_type The current type can then be subscripted like an array yielding values of type typelem A true array type is variable length typlen 1 but some fixed length typlen gt 0 types also have nonzero typelem for example name and oidvector If a fixed length type has a typelem then its internal representation must be N values of the typelem datatype with Ino other data Variable length array types have a header defined by the array subroutines typinput regproc Input function typoutput regproc Output function typreceive regproc unused typsend regproc unused 31 Chapter 3 System Catalogs
93. ry number to identify this group grolist int4 Ipg_shadow usesysid An array containing the ids of the users in this group 3 9 pg_index pg_index contains part of the information about indexes The rest is mostly in pg_class Table 3 9 pg_index Columns Name Type References Description indexrelid oid pp class od The oid of the pg_class entry for this index indrelid loid pp class od The oid of the pg_class entry for the table this index is for indproc regproc Ipg_proc oid The registered procedure if this is a functional index 18 Chapter 3 System Catalogs Name Type References Description indkey int2vector Ipg_attribute attnum This is a vector array of up to INDEX_MAX_KEYS values that indicate which table columns this index pertains to For example a value of 1 3 would mean that the first and the third column make up the index key iindclass loidvector Ipg_opclass oid For each column in the index key this contains a reference to the operator class to use See pg_opclass for details indisclustered bool unused indisunique Gool If true this is a unique index indisprimary bool If true this index represents the primary key of the table indisunique should always be true when this is true indreference oid unused indpred text
94. ss to trigger each null terminated Note pg_class reltriggers needs to match up with the entries in this table 3 21 pg_type This catalog stores information about datatypes Scalar types base types are created with CRE ATE TYPE A complex type is also created for each table in the database to represent the row structure of the table Table 3 21 pg_type Columns Name Type References Description typname name Data type name typowner int4 Ipg_shadow usesysid Owner creator of the type typlen int2 Length of the storage representation of the type 1 if variable length typprtlen int2 unused 29 Chapter 3 System Catalogs Name Type References Description typbyval bool typbyval determines whether internal routines pass a value of this type by value or by reference Only char short and int equivalent items can be passed by value so if the type is not 1 2 or 4 bytes long PostgreSQL does not have the option of passing by value and So typbyval had better be false Variable length types are always passed by reference Note that tt ypbyval can be false even if the length would allow pass by value this is currently true for type float4 for example typtype char typtype is b fora base type and c fora complex type i e a table s row type If typtype isc typrelid is the OID of the type s entry in DO Class typisdefined bool T
95. ssage response to SSLRequest from the server This would only occur if the server predates the addition of SSL support to PostgreSQL In this case the connection must be closed but the frontend may choose to open a fresh connection and proceed without requesting SSL An initial SSLRequest may also be used in a connection that is being opened to send a CancelRequest message While the protocol itself does not provide a way for the server to force SSL encryption the ad ministrator may configure the server to reject unencrypted sessions as a byproduct of authentication checking 4 3 Message Data Types This section describes the base data types used in messages Intn i An n bit integer in network byte order If i is specified it is the literal value Eg Int16 Int32 42 LimStringn s A character array of exactly n bytes interpreted as a null terminated string The zero byte is omitted if there is insufficient room If s is specified it is the literal value Eg LimString32 LimString64 user String s A conventional C null terminated string with no length limitation If s is specified it is the literal value Eg String String user Note There is no predefined limit on the length of a string that can be returned by the backend Good coding strategy for a frontend is to use an expandable buffer so that anything that fits in memory can be accepted If that s not feasible read the full string and discard trailing characters
96. stening for a ReadyForQuery message 35 Chapter 4 Frontend Backend Protocol ReadyForQuery Start up is completed The frontend may now issue query or function call messages ErrorResponse Start up failed The connection is closed after sending this message NoticeResponse A warning message has been issued The frontend should display the message but continue listening for ReadyForQuery or ErrorResponse The ReadyForQuery message is the same one that the backend will issue after each query cycle Depending on the coding needs of the frontend it is reasonable to consider ReadyForQuery as starting a query cycle and then BackendKeyData indicates successful conclusion of the start up phase or to consider ReadyForQuery as ending the start up phase and each subsequent query cycle 4 2 2 Query A Query cycle is initiated by the frontend sending a Query message to the backend The backend then sends one or more response messages depending on the contents of the query command string and finally a ReadyForQuery response message ReadyForQuery informs the frontend that it may safely send a new query or function call The possible response messages from the backend are CompletedResponse An SQL command completed normally CopyInResponse The backend is ready to copy data from the frontend to a table The frontend should then send a CopyDataRows message The backend will then respond with a CompletedResponse message with a tag of COPY
97. t 1989 M Stonebraker The case for partial indexes SIGMOD Record 18 4 Dec 1989 p 4 11 M Stonebraker L A Rowe and M Hirohama The implementation of POSTGRES Transactions on Knowledge and Data Engineering 2 1 IEEE March 1990 M Stonebraker A Jhingran J Goh and S Potamianos On Rules Procedures Caching and Views in Database Systems Proc ACM SIGMOD Conference on Management of Data June 1990 See oe http s2k ftp CS Berkeley EDU 8000 postgres papers ERL M87 13 pdf http simon cs cornell edu home praveen papers partindex de95 ps Z http s2k ftp CS Berkeley EDU 8000 postgres papers ERL M85 95 pdf http s2k ftp CS Berkeley EDU 8000 postgres papers ERL M87 06 pdf http s2k ftp CS Berkeley EDU 8000 postgres papers ERL M89 82 pdf http s2k ftp CS Berkeley EDU 8000 postgres papers ERL M89 17 pdf http s2k ftp CS Berkeley EDU 8000 postgres papers ERL M90 34 pdf 0 http s2k ftp CS Berkeley EDU 8000 postgres papers ERL M90 36 pdf 80
98. te Columns Name Type References Description laggname name Name of the aggregate function aggowner int4 Ipg_shadow usesysid Owner creator of the aggregate function aggtransfn regproc function Ipg_proc oid Transition function aggfinalfn regproc function Ipg_proc oid Final function laggbasetype oid Ipg_type oid The input datatype for this aggregate function laggtranstype oid Ipg_type oid The type of the aggregate function s internal transition state data aggfinaltype oid Ipg_type oid The type of the result agginitval text The initial value of the transition state This is a text field containing the initial value in its external string representation If the field is NULL the transition state value starts out NULL New aggregate functions are registered with the CREATE AGGREGATE command See the Pro grammer s Guide for more information about writing aggregate functions and the meaning of the transition functions etc An aggregate function is identified through name and argument type Hence aggname and aggbase type are the composite primary key 3 3 pg_attrdef This catalog stores column default values The main information about columns is stored in pg_attribute see below Only columns that explicitly specify a default value when the table is created or the column is added will have an entry here Table 3 3 pg_attrdef Columns Name Type Re
99. tem 7 xxxkkkKKK for SGML mode psgml setq sgml omittag t setq sgml shorttag t setq sgml minimize attributes nil setq sgml always quote attributes t setq sgml indent step 1 setq sgml indent data t setq sgml parent document nil setq sgml default dtd file reference ced setq sgml exposed tags nil setq sgml catalog files usr local share sgml catalog setq sgml ecat files nil autoload sgml mode psgml Major mode to edit SGML files t and in the same file add an entry for SGML into the existing definition for auto mode alist setq 15 http www lysator liu se projects about_psgml html 77 Appendix B Documentation auto mode alist sgml sgml mode Currently each SGML source file has the following block at the end of the file lt Keep this comment at the end of the file Local variables mode sgml sgml omittag t sgml shorttag t sgml minimize attributes nil sgml always quote attributes t sgml indent step 1 sgml indent data t sgml parent document nil sgml default dtd file reference ced sgml exposed tags nil sgml local catalogs usr lib sgml catalog sgml local ecat files nil End gt This will set up a number of editing mode parameters even if you do not set up your emacs file but it is a bit unfortunate since if you followed the installation instructions above then the
100. that don t fit into your fixed size buffer Byten c Exactly n bytes If c is specified it is the literal value Eg Byte Byte1 n 4 4 Message Formats This section describes the detailed format of each message Each can be sent by either a frontend F a backend B or both F amp B AsciiRow B Bytel D Identifies the message as an ASCII data row A prior RowDescription message defines the number of fields in the row and their data types 40 Chapter 4 Frontend Backend Protocol Byten A bit map with one bit for each field in the row The Ist field corresponds to bit 7 MSB of the Ist byte the 2nd field corresponds to bit 6 of the Ist byte the 8th field corresponds to bit 0 LSB of the Ist byte the 9th field corresponds to bit 7 of the 2nd byte and so on Each bit is set if the value of the corresponding field is not NULL If the number of fields is not a multiple of 8 the remainder of the last byte in the bit map is wasted Then for each field with a non NULL value there is the following Int32 Specifies the size of the value of the field including this size Byten Specifies the value of the field itself in ASCII characters n is the above size minus 4 There is no trailing zero byte in the field data the front end must add one if it wants one AuthenticationOk B Bytel R Identifies the message as an authentication request Int32 0 Specifies that the authentication was successful
101. the server 2 The parser stage checks the query transmitted by the application program client for correct syntax and creates a query tree 3 The rewrite system takes the query tree created by the parser stage and looks for any rules stored in the system catalogs to apply to the querytree and performs the transformations given in the rule bodies One application of the rewrite system is given in the realization of views Whenever a query against a view i e a virtual table is made the rewrite system rewrites the user s query to a query that accesses the base tables given in the view definition instead 4 The planner optimizer takes the rewritten querytree and creates a queryplan that will be the input to the executor It does so by first creating all possible paths leading to the same result For example if there is an index on a relation to be scanned there are two paths for the scan One possibility is a simple sequential scan and the other possibility is to use the index Next the cost for the execution of each plan is estimated and the cheapest plan is chosen and handed back 5 The executor recursively steps through the plan tree and retrieves tuples in the way represented by the plan The executor makes use of the storage system while scanning relations performs sorts and joins evaluates qualifications and finally hands back the tuples derived In the following sections we will cover every of the above listed items in more
102. tion prolang oid pg_language oid Implementation language or call interface of this function proisinh bool unused proistrusted bool not functional proiscachable bool Function returns same result for same input values 23 Chapter 3 System Catalogs Name Type References Description proisstrict bool Function returns null if any call argument is null In that case the function won t actually be called at all Functions that are not strict must be prepared to handle null inputs jpronargs int2 Number of arguments proretset bool Function returns a set ie multiple values of the specified datatype prorettype Ipg_type oid Data type of the return value 0 if the function does not return a value lproargtypes loidvector Ipg_type oid A vector with the data types of the function arguments lprobyte_pct int4 dead code properbyte_cpu int4 dead code lpropercall_cpu int4 dead code lprooutin_ratio int4 dead code lprosrce text This tells the function handler how to invoke the function It might be the actual source code of the function for interpreted languages a link symbol a file name or just about anything else depending on the implementation language call convention probin bytea Additional information about how to invoke the function Again the interpretation
103. tors pg_amproc access method support procedures pg_attrdef column default values pg_attribute table columns attributes fields pg_class tables indexes sequences relations pg_database databases within this database cluster pg_description descriptions or comments on database objects pg_group groups of database users Ipg_index additional index information pg_inherits table inheritance hierarchy pg_language languages for writing functions pg_largeobject large objects pg_listener asynchronous notification pg_opclass index access method operator classes pg_operator operators pg_proc functions and procedures pg_relcheck check constraints pg_rewrite query rewriter rules pg_shadow database users pg_statistic optimizer statistics pg_trigger triggers pg_type data types More detailed documentation of most catalogs follow below The catalogs that relate to index access methods are explained in the Programmer s Guide Chapter 3 System Catalogs 3 2 pg_aggregate pg_aggregate stores information about aggregate functions An aggregate function is a function that operates on a set of values typically one column from each row that matches a query condition and returns a single value computed from all these values Typical aggregate functions are sum count and max Table 3 2 pg_aggrega
104. ttr int2 The column this rule is for currently always zero to indicate the whole table is_instead bool True if the rule is an INSTEAD rule ev_qual text Expression tree in the form of a nodeToString representation for the rule s qualifying condition 25 Chapter 3 System Catalogs Name Type References Description ev_action text Query tree in the form of a nodeToString representation for the rule s action Note pg_class relhasrules must be true if a table has any rules in this catalog 3 18 pg shadow pg_shadow contains information about database users The name stems from the fact that this table should not be readable by the public since it contains passwords pg_user is a publicly readable view on pg_shadow that blanks out the password field The Administrator s Guide contains detailed information about user and permission management Because user identities are cluster wide pg_shadow is shared across all databases of a cluster there is only one copy of pg_shadow per cluster not one per database Table 3 18 pg_shadow Columns Name Type References Description lusename name User name lusesysid int4 User id arbitrary number used to reference this user lusecreatedb bool User may create databases lusetrace bool not used lusesuper bool User is a superuser lusecatupd bool User may update system catalogs Even a su
105. urned After a tuple has been handed back by the leftt ree of the MergeJoin the righttree is processed in the same way If both tuples are present the executor processes the MergeJoin node Whenever a new tuple from one of the subplans is needed a recursive call to the executor is performed to obtain it If a joined tuple could be created it is handed back and one complete processing of the plan tree has finished Now the described steps are performed once for every tuple until a NULL pointer is returned for the processing of the MergeJoin node indicating that we are finished 3 1 Overview Chapter 3 System Catalogs The system catalogs are the place where a relational database management system stores schema metadata such as information about tables and columns and internal bookkeeping information Post greSQL s system catalogs are regular tables You can drop and recreate the tables add columns insert and update values and severely mess up your system that way Normally one should not change the system catalogs by hand there are always SQL commands to do that For example CREATE DATABASE inserts a row into the pg_database catalog and actually creates the database on disk There are some exceptions for esoteric operations such as adding index access methods Table 3 1 System Catalogs Catalog Name Purpose pg aggregate aggregate functions pg_am index access methods pg_amop access method opera
106. wner of the database initially who created it encoding int4 Character multibyte encoding for this database datistemplate Gool If true then this database can be used in the TEMPLATE clause of CREATE DATABASE to create the new database as a clone of this one ddatallowconn bool If false then no one can connect to this database This is used to protect the template0 database from being altered datlastsysoid Last system OID in the database useful particularly to pg_dump 16 Chapter 3 System Catalogs Name Type References Description datvacuumxid xid All tuples inserted or deleted by transaction IDs before this one have been marked as known committed or known aborted in this database This is used to determine when commit log space can be recycled datfrozenxid xid All tuples inserted by transaction IDs before this one have been relabeled with a permanent frozen transaction ID in this database This is useful to check whether a database must be vacuumed soon to avoid transaction ID wraparound problems datpath text If the database is stored at an alternative location then this records the location It s either an environment variable name or an absolute path depending how it was entered 3 7 pg_description The pg_description table can store an optional description or comment for each database object
107. y this query but the actual length of this attribute is the length size of an oid At least this is the theory All this is probably quite broken these days lattalign char A copy of pg_type typalign of this column s type lattnotnull bool This represents a NOT NULL constraint It is possible to change this field to enable or disable the constraint atthasdef bool This column has a default value in which case there will be a corresponding entry in the pg_attrdef catalog that actually defines the value 3 5 pg_class pg_class catalogues tables and mostly everything else that has columns or is otherwise similar to a table This includes indexes but see also pg_index sequences views and some kinds of special relation Below when we mean all of these kinds of objects we speak of relations Not all fields are meaningful for all relation types Table 3 5 pg_class Columns Name Type References Description relname name Name of the table index view etc 13 Chapter 3 System Catalogs Name Type References Description reltype oid Ipg_type oid The OID of the data type that corresponds to this table if any zero for indexes which have no pg_type entry relowner int4 Ipg_shadow usesysid Owner of the relation relam oid Ipg_am oid If this is an index the access method used btree hash etc relfilenode oid Name of the on dis

PostgreSQL 7.2 Developer`s Guide

Contents

Download Pdf Manuals

Related Search

Related Contents